IntroductionWelcome| 00:00 |
Hello.
I'm Chris Nodder.
| | 00:04 |
Welcome to Foundations of UX Information
Architecture.
| | 00:10 |
In this course I'll be covering the steps
you need to follow in order to find out
| | 00:13 |
how your users think about the world, and
then how to turn that into the best
| | 00:17 |
possible navigation structure.
We'll discuss how to get information from
| | 00:23 |
customers using card sort and reverse sort
research, how to use that research to
| | 00:28 |
create a great information architecture,
and then how to use that information
| | 00:33 |
architecture to redesign navigation menus,
content classification, and page layout so
| | 00:38 |
that your site or application supports the
way your users think about the world.
| | 00:46 |
Improving the information architecture is
one of the most cost effective and most
| | 00:50 |
often overlooked ways of increasing your
user satisfaction.
| | 00:56 |
This course doesn't make any assumptions
about your background, although it's
| | 00:59 |
primarily aimed at people who work in, or
are learning about software design.
| | 01:05 |
Again.
This course doesn't make any assumptions
| | 01:08 |
about your background, although it's
primarily aimed at people (INAUDIBLE).
| | 01:11 |
Again.
This course doesn't make any assumptions
| | 01:17 |
about your background, although it's
primarily aimed at people who work in, or
| | 01:21 |
who (INAUDIBLE).
Again.
| | 01:23 |
This course doesn't make any assumptions
about your background, although it's
| | 01:30 |
primarily aimed at people who work in, or
who are learning about software design and development.
| | 01:37 |
Creating a suitable information
architecture is a major step in making a
| | 01:40 |
successful website or application.
Now, I want to help you learn how to make
| | 01:47 |
a site structure that shows your users
that you really understand them.
| | 01:50 |
So with that, let's get started.
| | 01:52 |
| | Collapse this transcript |
| Using the exercise files| 00:00 |
To help you through the information
architecture creation process, we've
| | 00:04 |
included two documents in the exercise
files directory that accompanies this course.
| | 00:08 |
Unlike most other lynda.com courses, you
don't need to have the documents open
| | 00:13 |
while you watch.
Instead, they'll be useful to you when
| | 00:16 |
you're planning your own study later on.
The documents give you generic templates
| | 00:20 |
that you can customize for your own use
when you run your own card sort studies.
| | 00:25 |
| | Collapse this transcript |
|
|
1. What Is Information Architecture?What is information architecture?| 00:00 |
People often use the words information
architecture to mean the menus on a
| | 00:04 |
website, but that's not really correct.
The menus are part of the information
| | 00:09 |
architecture, but they're only one part of
it.
| | 00:12 |
The Information Architecture Institute
says, we define information architecture
| | 00:17 |
as the art and science of organizing and
labeling websites, intranets, online
| | 00:22 |
communities and software To support
usability and findability.
| | 00:28 |
So, the term information architecture
refers to how you show visitors to your
| | 00:32 |
site or users of your application, the
content you have and the actions they can perform.
| | 00:37 |
As you can imagine, that encompasses your
menus, the items you put on the page Your
| | 00:42 |
site's structure and even the terminology
that you use to describe things.
| | 00:47 |
It obviously makes sense to show your
visitors or users this information in a
| | 00:51 |
way they normally think about it.
Then the navigation structure feels
| | 00:55 |
natural and blends into the background.
The content feels logically grouped.
| | 01:01 |
Searching, sorting, and filtering
information becomes second nature.
| | 01:04 |
The end result is that people can focus on
their tasks, not on finding their way around.
| | 01:10 |
It's unlikely that your site or product is
unique.
| | 01:14 |
There are probably several other
alternatives that people could use.
| | 01:18 |
Having a clear information architecture
that helps users easily complete their
| | 01:21 |
tasks, means they'll find your site or
product more usable.
| | 01:25 |
And so they'll be more likely to use it
than one of your competitors.
| | 01:29 |
On the other hand, if they can't find what
they're looking for, or if you don't
| | 01:33 |
present information the way they expect to
see it shown, people will abandon your
| | 01:37 |
site and move on to someone else's.
For that reason, having a clear and well
| | 01:42 |
thought out information architecture, is a
big competitive advantage.
| | 01:46 |
| | Collapse this transcript |
| Creating good information architecture| 00:00 |
The whole purpose of creating a good
information architecture is to help users
| | 00:04 |
navigate your site or product.
So it makes sense to include your users in
| | 00:09 |
the design process.
That means performing some user research
| | 00:13 |
tasks in order to find out how your users
categorize things.
| | 00:17 |
Once you have this information, you can
use it to find a creative solution.
| | 00:21 |
And then do some usability testing to make
sure your solution fits your user's needs.
| | 00:27 |
During this course, we'll show you how to
gather information from users, using a
| | 00:31 |
technique called card sorting.
Then, how to interpret that information.
| | 00:37 |
And turn it into an information
architecture.
| | 00:40 |
Next.
You'll test this architecture with some
| | 00:43 |
more users to verify it with a reverse
card sort.
| | 00:47 |
Before using it to create your site's
navigation structure and layout.
| | 00:50 |
At that point it's time to go back to your
user's again to use a Birdie Test your new
| | 00:56 |
site or application design.
The information you get from your users
| | 01:01 |
will help you to do the design work to
make your site understandable and navigable.
| | 01:07 |
Doing this research doesn't remove the
need for you to be creative, instead, it
| | 01:12 |
helps guide your creativity in the right
direction.
| | 01:14 |
Getting user input at each stage of the
process makes sure we stay on track to
| | 01:19 |
deliver the right solution that meets our
users needs.
| | 01:22 |
| | Collapse this transcript |
|
|
2. Research to Determine Information Architecture| Why do information architecture research? | Card sorting to determine information architecture| 00:00 |
In order to build a good navigation and
content structure for our site or
| | 00:04 |
application, we need to understand how our
users think about the topics that we cover.
| | 00:08 |
In other words, the concepts they use to
group things and the language they use to
| | 00:13 |
describe things.
To do this, we recruit representative
| | 00:17 |
users as participants in studies that are
designed specifically to reveal how they
| | 00:21 |
think about the concepts your site or
product uses.
| | 00:24 |
Typically, users aren't very good at
straight out telling you how things should
| | 00:29 |
be, so we use a technique called card
sorting to help them organize their
| | 00:33 |
thoughts and communicate what they want,
in a way that we can work with.
| | 00:38 |
Card sorting got its name, because the
first people who used the technique worked
| | 00:43 |
with three inch by five inch index cards.
The name stuck, and is still used even if
| | 00:49 |
the sort happens online and doesn't even
resemble the original format anymore.
| | 00:54 |
I'm going to explain the basic technique
first, and then later chapters will look
| | 00:58 |
at how you can apply that same technique
using computer based and online tools.
| | 01:03 |
In a card sorting exercise participants
sort a stack of index cards into groups
| | 01:08 |
while you watch and listen.
Each index card has a task written on it
| | 01:13 |
that people can perform on your site.
Participants read each card one by one and
| | 01:19 |
then place them to create groups of
similar tasks.
| | 01:22 |
After they've sorted the cards, we check
for groups that contain too many cards,
| | 01:26 |
and ask participants to break groups that
contain more than around ten cards into
| | 01:30 |
smaller groups.
When they're done, participants write a
| | 01:35 |
name for each group on a blank index card.
At this point, we can probe about areas
| | 01:41 |
that seemed to cause them problems during
the sort.
| | 01:43 |
In order to truly understand their reasons
for placing the cards in the groups they did.
| | 01:49 |
After we've done this same exercise with
enough participants, we can compare the
| | 01:53 |
groupings they made and the language they
used to describe the groups in order to
| | 01:57 |
come up with a set of categories that best
fits the way our users think about the
| | 02:01 |
information space.
This set of categories becomes our
| | 02:05 |
information architecture.
It really is almost as simple as it sounds.
| | 02:10 |
There are some important points to
consider when you're creating and running
| | 02:15 |
the card sort.
And analysis can be a bit tricky.
| | 02:18 |
But overall, the technique gives you a lot
of useful information for very little time investment.
| | 02:24 |
In the next chapter, I'll go through the
details of how to set up a card sort activity.
| | 02:28 |
| | Collapse this transcript |
|
|
3. Creating and Running a Paper Card SortFinding the information to use in a card sort| 00:00 |
We already mentioned that in a paper card
sort, participants sort a stack of index
| | 00:05 |
cards into groups while you watch and
listen.
| | 00:08 |
We also said that each index card has a
task written on it that people can perform
| | 00:15 |
on your site.
So, the first thing you need to think
| | 00:17 |
about is what are the key tasks that
people will want to perform on your site.
| | 00:23 |
That could be activities that they want to
do, information they'll want to find, help
| | 00:29 |
that they'll need, products that they'll
search for, or, services that you offer
| | 00:34 |
that would benefit them.
Most often, you'll have an existing site
| | 00:38 |
or product that you're trying to improve.
That makes you job easier, because you can
| | 00:44 |
make a list of tasks by looking at the
content and tasks that the current product supports.
| | 00:50 |
Be careful though, just because you offer
something doesn't mean it's what people
| | 00:54 |
want and you maybe missing a thing that
you don't realize people need.
| | 00:58 |
If your building a brand new product or
site, write the list of things you want to have.
| | 01:04 |
Get ideas by looking at some other
examples that already exist in a similar space.
| | 01:10 |
If you're designing an application, as
well as the content, you want to list out
| | 01:14 |
the actions people can perform with the
application, that is, their tasks.
| | 01:19 |
I suggest you check out Patrick Nichols
course on content strategy if you want
| | 01:23 |
more detail on how to perform a content
audit and a gap analysis.
| | 01:28 |
Start narrowing things down by identifying
a set of items that exist at a similar level.
| | 01:34 |
It's hard for people to sort cards when
there's already an implicit hierarchy in
| | 01:38 |
the different cards.
So, using a florist's site as an example,
| | 01:43 |
we'd be better off choosing items like
mixed arrangements, caring for orchids,
| | 01:49 |
and information about the florist.
Which are at an equivalent level of hierarchy.
| | 01:53 |
Even though they're from different
sections of the site then we would to
| | 01:56 |
choose arrangements, orchids and then
Oncidium which is a type of orchid because
| | 02:03 |
those three items are at different levels
of detail.
| | 02:06 |
Oncidium orchids are in the orchid
arrangement which is an example of an arrangement.
| | 02:12 |
So they can't easily be compared to each
other.
| | 02:15 |
If you choose items at different levels,
it's likely to drive people towards
| | 02:19 |
creating groups based on the top level
items.
| | 02:23 |
As a guideline, it's good to end up
working at a level of content that can
| | 02:27 |
give you around 30 to 100 cards.
That probably involves starting from a
| | 02:32 |
list of about 200 items.
For a big site, that might mean choosing a
| | 02:37 |
level quite far up in the hierarchy.
For a smaller site, that could be
| | 02:42 |
individual items.
For a really big site, you might have to
| | 02:46 |
run several different card sorts.
One at the higher level to find the
| | 02:50 |
overall architecture, then individual
sorts which represents the participants
| | 02:55 |
for each individual section of the
architecture.
| | 02:58 |
That's especially true when different
types of people are likely to use
| | 03:01 |
different parts of the site.
When you have a list of tasks, the next
| | 03:06 |
step is to work at how to properly phrase
them for the card sort exercise.
| | 03:11 |
That's what we'll cover next.
| | 03:12 |
| | Collapse this transcript |
| Deciding what goes on the cards| 00:00 |
There isn't much space on each card.
And we don't want our card sort
| | 00:04 |
participants to have to read too much
text.
| | 00:07 |
We also don't want the card information to
lead people to a certain answer.
| | 00:12 |
Take the list of items you've created; and
make sure that every item is phrased
| | 00:16 |
either in terms of the content, or in
terms of the tasks that people can perform.
| | 00:20 |
But not both.
That sounds hard.
| | 00:23 |
But as an example, if you choose to go the
tasks route, you can convert content into
| | 00:29 |
tasks quite easily.
For instance, content on caring for
| | 00:33 |
orchids, can become a task to find out how
to care for an orchid.
| | 00:37 |
It's harder to turn tasks into content,
which is why I tend to write my card sort
| | 00:42 |
card as tasks.
This also has an advantage later when we
| | 00:47 |
look at reverse sorting.
because the reverse sort activity uses
| | 00:51 |
these same tasks as its input.
It's important to phrase the tasks in
| | 00:56 |
users' terms.
What I mean by that is that tasks should
| | 01:00 |
be realistic and believable.
If we were writing tasks for a card sort
| | 01:05 |
on a florist's site, we shouldn't just
write.
| | 01:07 |
Find an article on orchids.
That's not particularly realistic.
| | 01:13 |
It's unlikely a user would ever set out to
do that.
| | 01:16 |
Instead, you might write, "You were just
given a Dendrobium orchid.
| | 01:20 |
How would you make sure it lives a long
time?" This is specific and realistic.
| | 01:28 |
So, your list of tasks needs to be user
centric.
| | 01:31 |
Each task also needs to be active.
Although you are ultimately interested in
| | 01:35 |
where to put the content on your side,
thus know how users think.
| | 01:39 |
Instead they think about how they think
about how they are going to achieve their goals.
| | 01:43 |
What really matters to them is where they
think you should go to get answers.
| | 01:50 |
In other words where they go to achieve
tasks.
| | 01:54 |
Try to keep each task as short as possible
without being ambiguous.
| | 01:59 |
Once you've got your file list, read it
through and make sure you aren't
| | 02:02 |
inadvertantly guiding people in a certain
direction.
| | 02:06 |
By this, I mean repeating a certain
phrasing on several of the cards, like
| | 02:10 |
find out how to, or get information about.
If several cards have that phrasing, then
| | 02:17 |
casual participants may just place them
all in that how to group without giving
| | 02:22 |
them anymore detailed thought.
The same thing can happen if you overuse a
| | 02:27 |
certain word.
Participants will just group every
| | 02:30 |
occurrence of that word.
You need a list of around 30 to 50 tasks
| | 02:36 |
for a good card sort exercise.
If you have less than 30, it's hard for
| | 02:41 |
participants to make good groupings.
If you have more than 50, the card sort
| | 02:46 |
can take a long time and participants
might start losing interest.
| | 02:51 |
The tasks should be representative of all
the areas of your site or product.
| | 02:55 |
So for larger sites with lots of content
categories, I've used 100 tasks in the past.
| | 03:01 |
But the more tasks you use, the harder the
sort is for participants to perform, and
| | 03:05 |
the harder the analysis is for you.
| | 03:07 |
| | Collapse this transcript |
| Making the cards| 00:00 |
Once you have your list of tasks, it's
time to make them into cards for the
| | 00:04 |
sorting exercise.
You can do that in several ways.
| | 00:09 |
The lowest tech way to make cards is to
write them out by hand.
| | 00:13 |
You can make a set of 30 three inch by
five inch cards quite quickly.
| | 00:18 |
This is normally sufficient for most
sorts.
| | 00:21 |
Handwriting your cards does have a few
drawbacks.
| | 00:24 |
The first issue is that creating more sets
of cards takes a long time.
| | 00:28 |
Often, you might want to conduct a series
of cards so it's one after the other, so
| | 00:33 |
you'll want to leave them bundled up into
groupings that participates used, until
| | 00:37 |
you have time to analyze them later.
Or, you might have two people running card
| | 00:42 |
sorts at the same time.
That means you'll need two or more sets of cards.
| | 00:48 |
Writing out a new set of cards takes as
much time as the first set took.
| | 00:53 |
The other issue is how legible the cards
are.
| | 00:56 |
If you have neat writing and if you write
large enough using a thick pen, you'll
| | 00:59 |
probably be okay.
However, if your writing is messy, it will
| | 01:04 |
make things harder for your participants.
That's why I often print my cards out.
| | 01:10 |
There are a few different ways of doing
this.
| | 01:13 |
One is to print the tasks on sticky labels
and then stick each label on an index card.
| | 01:18 |
I find that the sheets that give you two
columns of five labels on one page work
| | 01:24 |
well for this.
Another way is to use heavyweight paper,
| | 01:28 |
and print directly onto a regular letter
or A4 size sheet, then you can cut the
| | 01:33 |
sheet into four cards.
You'll find that 65 pound card stock,
| | 01:38 |
that's 175 grams per square meter, will go
through most laser printers without any trouble.
| | 01:44 |
And it's stiff enough to use for our
purposes.
| | 01:48 |
You can even buy card stock designed for
laser printers that already has
| | 01:51 |
perforations so you can easily tear it
into four cards.
| | 01:55 |
If you have the money for the special
card, that's the easiest technique to use.
| | 02:01 |
Whether I print my tasks or hand write
them, I always include a reference number
| | 02:05 |
on each card.
That, plus the names that people give to
| | 02:09 |
each group of cards will be the
information that you need for analysis.
| | 02:14 |
The reference number should be sequential
from one to the number of cards you have.
| | 02:19 |
But, be careful not to group tasks that
are similar to each other together in your
| | 02:23 |
list, so they don't have sequential
numbers.
| | 02:26 |
You don't want participants to start
grouping your tasks based on the numbers
| | 02:30 |
they see written next to each one.
So, you can use any technique, from
| | 02:36 |
handwriting your tasks, through to laser
printing onto special per-perforated paper.
| | 02:41 |
Remember though, these cards are only a
means to an end.
| | 02:44 |
They don't have to be beautifully
prepared, they just have to be good enough
| | 02:47 |
for people to use in a card sort exercise.
| | 02:50 |
| | Collapse this transcript |
| Recruiting participants| 00:00 |
You'll need to find people who are
representative users of your site or product.
| | 00:05 |
If you're creating the information
architecture for your company's intranet
| | 00:08 |
or an internal tool, then you must find
people from the organization who would be
| | 00:12 |
likely users.
If you're creating a public facing
| | 00:16 |
website, you need to find people who would
be your site's customers or visitors.
| | 00:22 |
I covered the steps you'll need to cover
to recruit people in a lot of detail in my
| | 00:26 |
course on usability testing.
I suggest you watch the relevant chapters
| | 00:30 |
of that course before you start your
recruiting effort.
| | 00:34 |
At a minimum you'll need to find
sufficient people.
| | 00:37 |
Check that they meet your criteria.
Schedule times to meet with them, either
| | 00:41 |
at your place of work or at their
location, and convince them to turn up by
| | 00:45 |
paying them a suitable gratuity for their
time.
| | 00:49 |
15 participants should give you enough
data to have sufficient confidence in the results.
| | 00:54 |
Obviously, if you have different types of
users, you'll want to have this number of
| | 00:58 |
participants for each user type.
Because you'll need to see whether they
| | 01:02 |
think of the structure the same way or
not.
| | 01:05 |
For instance, a hardware store website
might have two distinct audiences.
| | 01:10 |
People who are going to do some
do-it-yourself home decorating could be
| | 01:14 |
very different from professional
contractors.
| | 01:16 |
One beauty of paper based card sorting is
that it doesn't take much setup.
| | 01:22 |
So, you can travel to your participants'
locations rather than bringing them in to
| | 01:27 |
your offices to run the sort if you'd
like.
| | 01:30 |
Just be sure that you have a large table
to run the sort on, in a location without
| | 01:35 |
any disturbances.
Also, make sure you run the sort indoors.
| | 01:39 |
There's nothing like a gust of wind to
ruin a card sort activity.
| | 01:43 |
| | Collapse this transcript |
| Running the session| 00:00 |
I suggest that you start by doing card
sorts one on one with your participant.
| | 00:05 |
In other words, you sit and watch as one
participant completes the sort.
| | 00:12 |
This allows you to really focus on the
groupings they create and listen to their justifications.
| | 00:17 |
Once you've run and analyzed a few
sessions, this way you might decide to do
| | 00:21 |
sorts with a group of two or three
participants working together at the same time.
| | 00:26 |
A group will probably talk more between
themselves, which can be useful
| | 00:31 |
information, but can also be too hard to
follow when you're just starting out.
| | 00:36 |
Before you even start with participant
sessions, do a run through with a friend
| | 00:40 |
or colleague to check for errors, typos,
or misconceptions that they have.
| | 00:45 |
I can't emphasize this enough.
There will be mistakes or confusion, and
| | 00:49 |
it's better to catch this with a friendly
helper than to have to throw out real user data.
| | 00:55 |
During a run through of the whole process,
we'll also make you more confident when
| | 00:58 |
you're working with real participants.
As the session moderator, your job is to
| | 01:03 |
make sure that participants feel
comfortable and understand what they're
| | 01:06 |
supposed to be doing, and then just sit
and watch quietly.
| | 01:12 |
I like it to read my introduction and
instructions from a moderator's script, so
| | 01:16 |
that I say the same thing to each
participant.
| | 01:19 |
Then I hand them the stack of cards and
let them get on with it.
| | 01:24 |
Getting participants to think out loud
will you give a wealth of additional information.
| | 01:30 |
When participants have doubts about where
to place an item their comments and the
| | 01:34 |
vocabularly they use can help you to work
out what alternative arrangements might be feasible.
| | 01:39 |
Similarly when they're very sure about an
item, it can give you an increased
| | 01:45 |
confidence about its placement.
Have a notepad handy to write down what
| | 01:49 |
each participant says, because the
terminology they use and the reasons they
| | 01:54 |
give can be very valuable to you as you do
your analysis and create your information architecture.
| | 02:01 |
My usability testing course gives you the
full details on moderating a session using
| | 02:05 |
the Think Out Loud Protocol.
And we've also included a sample card sort
| | 02:09 |
moderator script in the exercise files for
this course.
| | 02:12 |
I start by getting the participants to
spread all the cards out on the table.
| | 02:20 |
That lets them get familiar with the types
of information the cards contain, which in
| | 02:24 |
turn, means they can create better groups.
Then, they start grouping the cards.
| | 02:30 |
Once the participant has placed all the
cards into groups, check for groups that
| | 02:35 |
contain too many cards, perhaps ten or
more.
| | 02:37 |
And ask the participant to break that
stack of cards into smaller groups.
| | 02:43 |
We do this because otherwise, the groups
they create won't have sufficient differentiation.
| | 02:48 |
You might end up with a group called
products, when you really need to know how
| | 02:52 |
people distinguish between your different
product categories.
| | 02:56 |
It's also good to ask participants to see
if they can combine groups of just two
| | 03:00 |
cards with other groupings.
Sometimes that just won't be possible, but
| | 03:05 |
such small groups really don't help us
with analysis.
| | 03:09 |
If participants feel really strongly that
a certain card doesn't group well with any
| | 03:14 |
other cards that's okay.
However find out in detail why they feel
| | 03:19 |
that way then you'll know whether it was
because of the set of cards you used.
| | 03:24 |
For instance not having any other similar
cards or whether it was because of their
| | 03:28 |
mental concepts of the site.
Now, it's time for participants to tell
| | 03:34 |
you what they think each group should be
called.
| | 03:37 |
The best way to do this is normally to
give the participants some blank index
| | 03:41 |
cards and a pen.
Then, they can write each group name down
| | 03:45 |
on a card, and place the card on top of
each group.
| | 03:50 |
Getting participants to physically write
their names down and place the cards,
| | 03:54 |
means that they consider all the groupings
in combination.
| | 03:56 |
That way, the names they've given are more
likely to be unique and descriptive.
| | 04:02 |
I've even had participants who choose to
re sort some cards after they've tried
| | 04:06 |
unsuccessfully to give names to the groups
they've created.
| | 04:09 |
That's just fine.
At the end of the session you may have
| | 04:14 |
some questions about the groupings that a
participant chose, I'd suggest the
| | 04:18 |
following 3 questions as a good starting
point to get the information you need.
| | 04:22 |
The first is, what is the overall pattern
here?
| | 04:28 |
Asking this question, let's us understand
the stretcher they used as a basis for the groups.
| | 04:32 |
Next, I'd ask, were there any groups
difficult to create?
| | 04:38 |
This suggests items that may not fit well
together.
| | 04:42 |
If there are any, then ask them why those
were difficult to create.
| | 04:45 |
Then I'd ask were any cards difficult to
put into a group, and if so why.
| | 04:53 |
This suggests items that might not be
structurally related to each other.
| | 04:57 |
Be careful asking any other types of
questions.
| | 05:01 |
Remember asking leading questions or ones
that suggest that the participant choose
| | 05:05 |
the wrong grouping will just make your
participant feel bad and won't help you at all.
| | 05:10 |
Instead ask mutual questions like: can you
tell me more about this grouping or can
| | 05:16 |
you explain what you mean by this category
name.
| | 05:18 |
Finally thank the participant.
Give them their gratuity, and answer any
| | 05:25 |
questions they may have.
After they're gone, record the numbers of
| | 05:29 |
all the cards in each group on the card
the participant used to label the group.
| | 05:35 |
That way, you have a written record of the
card sort.
| | 05:38 |
Now, you can gather up all the card sort
cards, shuffle them so they're in a random
| | 05:43 |
order, and leave them ready for the next
participant.
| | 05:46 |
| | Collapse this transcript |
| Recording participants' answers| 00:00 |
After each participant has completed the
card sort exercise, you'll need to record
| | 00:04 |
the groupings they used in a spreadsheet.
Use the participant name as the
| | 00:09 |
identifier, then the name they gave to the
first group, followed by the reference
| | 00:14 |
numbers from the card they put in that
group.
| | 00:17 |
Repeat this down the page for each group
in turn.
| | 00:21 |
If some of your participants created
hierarchies of groups, say cut flowers and
| | 00:26 |
live flowers, underneath the heading of
flowers, just ignore the hierarchy for the moment.
| | 00:32 |
Make a separate group for each label that
was written.
| | 00:35 |
Keep a note of the hierarchies the
participants created, though.
| | 00:38 |
That can be useful later in the process.
Here's a little trick.
| | 00:43 |
You can check that you entered the data
correctly by doing the sum of the column.
| | 00:49 |
The sum of the numbers in the column
should be the same as the sum of all the
| | 00:53 |
numbers from one to the number of cards
you have.
| | 00:57 |
The formula is really simple, number of
cards mutiplied by number of cards plus one.
| | 01:02 |
Divided by two.
So if you had 30 cards, it would be 30
| | 01:07 |
times 31 divided by two.
After you've run all of your participants,
| | 01:12 |
you'll end up with a spreadsheet full of
group names, each name containing a set of
| | 01:16 |
card reference numbers.
This is the data that you'll need for your analysis.
| | 01:21 |
Always keep a copy of this raw data even
when you start manipulating it to get the
| | 01:25 |
answers you need.
You might well find that if the first
| | 01:28 |
version of your information architecture
design doesn't work as well as you'd
| | 01:31 |
hoped, you'll want to come back to the
original information to check things out.
| | 01:35 |
| | Collapse this transcript |
|
|
4. Analyzing a Paper Card SortGetting from cards to knowledge| 00:00 |
Each participant is likely to have sorted
the cards into slightly different groups,
| | 00:05 |
and quote those groups slightly different
things.
| | 00:07 |
Still, hopefully just from watching the
card sort sessions you'll already have
| | 00:12 |
noticed some general agreement between
participants.
| | 00:15 |
Or the emotions of maybe two separate ways
of looking to a site's contents or tasks.
| | 00:20 |
Now, we want to get a bit more rigorous
with our analysis.
| | 00:26 |
We already talked about capturing the raw
data in an Excel file.
| | 00:30 |
And turning into a grid of participant
card names for each task.
| | 00:35 |
We could probably have recorded our sort
data directly into this grid format after
| | 00:39 |
each card sort session.
But it's really useful to have both views,
| | 00:44 |
with the data sorted by group name and
also sorted by task name.
| | 00:48 |
Sorting by group name lets you quickly
tell how many groups each participant
| | 00:53 |
created, and how large each group was.
Sorting by task name let's you know how
| | 01:00 |
many groups or group names participants
placed each task into.
| | 01:04 |
Now we have the data in a more compact
format, it's time to rationalize those
| | 01:08 |
group names.
It's likely that several participants used
| | 01:12 |
similar names for groups, like maybe about
us, or company information, or even just
| | 01:17 |
the company name.
If those groupings tend to contain similar
| | 01:20 |
cards, it's fair to give them all the same
name.
| | 01:23 |
This isn't necessarily the final name
we'll give to this category.
| | 01:26 |
But it's a good way of reducing the range
of different group names to a common set.
| | 01:31 |
Make a copy of your original data in a new
sheet to the spreadsheet.
| | 01:35 |
Then replace all the original category
names the participants used.
| | 01:38 |
With the smallest set of standardized
names.
| | 01:41 |
Whenever you can, you want to draw out the
underlined contents of the category and
| | 01:44 |
the name you give it.
Look for synonyms like check out at the
| | 01:49 |
card, or basket, or common nouns and verbs
that participants used.
| | 01:56 |
In our data, participants 6, 7, and 8 have
given the first task group labels of
| | 02:03 |
general store information, store FAQs, and
information about the store.
| | 02:08 |
Those 3 can proabbly be combined into one
standard label.
| | 02:13 |
Choose the most frequently used label, or
the one that best fits in a family of
| | 02:17 |
labels with other that people have chosen.
For instance if participants tended to
| | 02:21 |
give action-based labels choose one that
is verb-based.
| | 02:25 |
If instead participants tended to be
descriptive choose a label that is noun-based.
| | 02:31 |
Most of the time you can use search and
replace to find all instances of a word
| | 02:32 |
and replace it with your standardized
category name, but before you do that,
| | 02:33 |
just make sure that all of your
participants use the term that you're
| | 02:34 |
replacing to mean the same thing.
For instance, it's possible that two
| | 02:35 |
people used the term support, but one
applied it just to questions about
| | 02:35 |
shipping, whereas the other applied it to
questions about products.
| | 02:36 |
In that instance, we would want to create
two standardized labels that actually
| | 02:37 |
differentiate those groups.
Like shipping support and frequently asked questions.
| | 03:07 |
Here you can see the results of my
standardized groupings.
| | 03:11 |
Once you have your reduced set of catgory
names it's time to work out how many
| | 03:15 |
particpants used that category in their
groupings.
| | 03:18 |
You can do this in a standardization grid.
On another new Excel work sheet type each
| | 03:25 |
of your standardized category names across
the top, and then list all your tasks down
| | 03:29 |
the side.
For each participant check off the
| | 03:33 |
standardized catogory name they placed
each task into.
| | 03:35 |
You'll end up with a tally of matches for
each cell.
| | 03:39 |
Now if you apply a conditional formatting
rule to show more color with more matches,
| | 03:45 |
you can easily see which categories are
chosen most frequently for each task.
| | 03:50 |
Here, for instance, you can see that tasks
12 and 13 get a corsage for a prom and get
| | 03:55 |
Valentine's day roses, were both very
frequently grouped into the special events category.
| | 04:00 |
The process of working with your data and
creating the standardized group names from
| | 04:04 |
participant's names will make you very
familiar with the types of groupings that
| | 04:08 |
participants chose.
There really is no short cut, no special
| | 04:12 |
algorithm to apply, in order to get this
familiarity.
| | 04:15 |
Eyeballing the data this way is the
easiest but least precise technique for
| | 04:20 |
creating a information architecture.
This will give you the general idea about
| | 04:25 |
the groupings that participants used and
the type of contents they expect in each group.
| | 04:30 |
Sometimes this is sufficient as the basis
for creating your information architecture.
| | 04:35 |
But most likely, you'll want a more robust
understanding of which items were
| | 04:38 |
consistently placed together by different
groups of participants and the
| | 04:41 |
hierarchical structure of those groups.
Luckily, there's software that helps us do
| | 04:45 |
deep analysis of the clusters of results
between participants the two techniques,
| | 04:51 |
eye balling the data and using cluster
analysis, compliment each other.
| | 04:55 |
And the next section will describe the
visual methods you can use.
| | 05:00 |
We'll discuss the software-based cluster
analysis tools when we talk about computer
| | 05:04 |
based sorts in the next chapter.
| | 05:05 |
| | Collapse this transcript |
| Eyeball analysis of your data| 00:00 |
Just working with the data, you'll start
to gain insights into how people grouped
| | 00:04 |
the cards into a structure.
By putting the information into certain
| | 00:07 |
formats, standardizing the names and
creating a standardization grid, you'll
| | 00:11 |
start to see patterns.
You won't be coming into the analysis as a
| | 00:15 |
completely neutral observer, and that's
okay.
| | 00:19 |
You probably have your own ideas about how
people are likely to sort the cards, and
| | 00:23 |
what labels they might give.
What's important is to be aware of
| | 00:26 |
anything that confirms or goes counter to
your preconceptions.
| | 00:30 |
Be careful not to brush aside unusual
data.
| | 00:33 |
It's this type of surprising information
that can give you great insights into how
| | 00:37 |
your users think differently than you do.
There are certain questions that you
| | 00:42 |
should ask to keep yourself honest.
Do people tend to create the same groups,
| | 00:46 |
or differrent ones?
If they create the same groups, then
| | 00:49 |
there's likely to be good agreement about
the final heirarchy.
| | 00:52 |
If differernt participants came up with
very different groups, are there patterns,
| | 00:56 |
like two separate ways of arranging items.
Often, by just comparing the groups that
| | 01:01 |
participants created, you'll see two or
more distinct patterns.
| | 01:05 |
It's interesting to see what
characteristics the participants who used
| | 01:08 |
a certain pattern share.
Is it experienced versus novice users?
| | 01:13 |
Customers versus suppliers?
Technicians versus sales people?
| | 01:19 |
Did people tend to use the same labels, or
different ones?
| | 01:23 |
If there's little variation, there's
likely to be good agreement about the
| | 01:26 |
final information architecture labels.
Again, if there's some varation, see if
| | 01:31 |
you can find out the reasons.
Are there two different groups of users
| | 01:35 |
who've got a different way of describing
things?
| | 01:38 |
It's also interesting to see what kind of
language participants used.
| | 01:41 |
Was it formal or informal.
Precise or general.
| | 01:44 |
Friendly or stern.
This is a good hint about how they view
| | 01:49 |
the subject matter and the company
culture.
| | 01:51 |
It makes sense to use a style of language
that mirrors participants' choices in your
| | 01:56 |
final structure.
Because that's likely to resonate better
| | 01:59 |
with your audience.
How did participants organize information?
| | 02:03 |
Did they group it by tasks?
By its intended audience?
| | 02:07 |
Or by topic?
Did they sort items by price, customer
| | 02:10 |
segment, style or brand?
Obviously, in your final design you can
| | 02:15 |
use searching, sorting and filtering tools
to give people any combination of these things.
| | 02:19 |
But it's helpful to know what your
participants consider to be the default
| | 02:22 |
view or what different options they
expected to have.
| | 02:27 |
Use the standardization grid to see which
cards are most frequently put in each group.
| | 02:32 |
Notice how here, there's good agreement
that the standard group label, store
| | 02:36 |
information, is where the where's the
store task which is number 17, should be placed.
| | 02:41 |
Also, look at cards that are put in many
different groups.
| | 02:45 |
For instance, there's much less agreement
about where the air plots task, number
| | 02:49 |
four, should be placed.
Some participants put it in live plants
| | 02:54 |
whereas others put it in plant care or
even ecological or school project.
| | 02:59 |
You have to work out whether this was
because your wording on the card was
| | 03:02 |
ambigious, whether the task was too
different from other tasks to be grouped
| | 03:06 |
with them, or whether the item really
could be sorted many different ways.
| | 03:11 |
In this situation there were no other
tasks like the air plants one, so it's
| | 03:15 |
unlikely the participants could have
placed it in the good group.
| | 03:18 |
That's my fault as the card sort creator.
If however it really was possible that it
| | 03:23 |
could easily set in several categories
you'll have to chose one but in your site
| | 03:28 |
design you'll probably need several see
also or related links style links from
| | 03:32 |
other place on the site.
Now go back to your notes.
| | 03:37 |
What did people say about the sort while
they were performing it?
| | 03:41 |
They might have used different words to
describe the content than they ended up
| | 03:44 |
writing on their group labels.
They may also have described their
| | 03:47 |
rationale for sorting a specific way.
Either by saying why two cards don't
| | 03:52 |
belong together.
Or by making comparisons between two or
| | 03:55 |
more groups they formed.
There's much to be learned from just
| | 03:59 |
looking at your data.
I always start by browsing the
| | 04:02 |
standardization grid, and rereading my
notes.
| | 04:04 |
Often that's enough to suggest an initial
information architecture.
| | 04:09 |
However, it's also sometimes useful to run
some statistical analysis of the same data
| | 04:14 |
using tools that perform cluster analysis
on participants groupings.
| | 04:18 |
We'll cover cluster analysis in the next
chapter when we talk about computer based
| | 04:22 |
card sorting.
| | 04:23 |
| | Collapse this transcript |
|
|
5. Running and Analyzing a Computer-Based Card SortAccessing remote users with online sorts| 00:00 |
Call me old-fashioned, but there's
something very powerful about watching
| | 00:03 |
participants place physical cards in
groups on a table, while you listen to
| | 00:07 |
their reasons for creating a particular
structure.
| | 00:09 |
However, paper-based sorts do end up
creating some data entry and analysis
| | 00:15 |
issues, and they aren't always practical
to run.
| | 00:18 |
There are software alternatives that
emulate a paper-based card sort.
| | 00:22 |
They capture all the data for you, and
they often spit out cluster analysis visualizations.
| | 00:27 |
Without any extra work on your behalf.
I would suggest checking into desktop
| | 00:32 |
based card sort apps.
Like XSORT, for the MAC, or USXORT for
| | 00:37 |
Windows, which are both free to use.
If you don't have physical access to
| | 00:42 |
participants, you could just use online
conferencing to share one of the desktop
| | 00:47 |
based card sort apps with a participant.
But there are also purpose built online tools.
| | 00:52 |
These online tools are also really useful
if you're tight on time, and need to run
| | 00:57 |
multiple sorts at once, or if you can't be
in the same location as your participants.
| | 01:03 |
Optimal Workshop's Optimal Sort is my
current favorite online tool because of
| | 01:07 |
its built-in data analysis and potential
to export to spreadsheets for further tweaking.
| | 01:12 |
Others you might want to consider are
Websort by uxpunk, and userzoon.
| | 01:19 |
All three are for trial accounts with
limited numbers of cards and participants.
| | 01:24 |
These are okay for testing out the
servers, but for a serious sort, you're
| | 01:27 |
probably going to have to put down some
money.
| | 01:29 |
Optimal Workshop's OptimalSort tool is
subscription based.
| | 01:34 |
Costing $109 per month for unlimited
studies and participants.
| | 01:38 |
UX Punk's Websort is $149 for one study
with up to 100 participants.
| | 01:46 |
Or $299 for three studies.
User zoom bundles several of their
| | 01:51 |
services together including card sorting
for $9,000 per year.
| | 01:56 |
Whether that's worth it to you will depend
on how frequently you plan on running card
| | 02:00 |
sorts or using the other tools in their
suite.
| | 02:04 |
Online tools typically let participants do
the sort individually in their own time.
| | 02:08 |
The downside is that you can't listen in
unless you've arranged to call the
| | 02:13 |
participant at the same time.
So you can't easily find out why people
| | 02:17 |
placed cards where they did.
Understanding their motivation is often
| | 02:21 |
really helpful when you're converting the
IA into a navigation structure.
| | 02:25 |
Another issue is that you'll probably not
be able to include so many cards.
| | 02:30 |
Screen real estate becomes a big issue,
and remote users tend to be less motivated.
| | 02:35 |
50 cards is about the maximum you can
safely expect someone to sort online.
| | 02:40 |
one other important point that might stop
you from using the web based tools in face
| | 02:45 |
to face situation, is that you obviously
need an internet connection.
| | 02:49 |
But there are benefits, as well.
You get fast, easy access to more
| | 02:53 |
participants, potentially across multiple
time zones.
| | 02:57 |
You can record other information about
each participant using built-in survey tools.
| | 03:02 |
You get real-time status updates on how
many people have completed the sort, and
| | 03:06 |
the tools provide data interpretation
automatically.
| | 03:10 |
The reason I like Optimal Workshops tool
is that it gives a lot of features for the
| | 03:14 |
price, and even after your subscription
expires, you can start access and analysis
| | 03:19 |
any data you collected.
So, you aren't held ransom.
| | 03:22 |
It's also iPad friendly, which is a big
deal if you want to use it for inversion sorts.
| | 03:29 |
To be clear, I'm not being paid to say
that.
| | 03:31 |
I'm just a happy customer.
If none of the tools I've mentioned seem
| | 03:35 |
right for you after you've checked them
out, there are many other vendors out there.
| | 03:38 |
So it should be easy to find one with the
features you need at a suitable price point.
| | 03:44 |
I still suggest that you watch my
demonstration on how OptimalSort works, so
| | 03:47 |
that you can make a good comparison with
other products.
| | 03:50 |
| | Collapse this transcript |
| Setting up a card sort using OptimalSort| 00:00 |
Let's go through the stages of setting up
a card sort using Optimal Workshop's
| | 00:05 |
OptimalSort product.
You'll find that most tools follow the
| | 00:08 |
same basic steps, although some make it
easier than others.
| | 00:12 |
After you've signed up and signed in, if
this is your first sort, you'll see a
| | 00:17 |
screen telling you what the tool does.
As you can see, with a free account you
| | 00:22 |
can run a proper card sort, but you're
limited to 30 cards and ten participants.
| | 00:28 |
I like the confidence of having at least
15 participants and often, if I have a
| | 00:33 |
disparate user base, I'll be testing 15
participants from each of several
| | 00:37 |
different user types.
So, maybe 45 participants total.
| | 00:41 |
For that reason, I use the paid account.
The analysis tools are the same, so if you
| | 00:47 |
aren't in a position to pay, follow along
anyway using the free account.
| | 00:51 |
However, I'm going to click the Upgrade
Now button to give them my money.
| | 00:57 |
Now I'm back at the same screen after
paying for a one month subscription.
| | 01:01 |
Like I said, the process is the same for
paid or free accounts.
| | 01:06 |
Let's start by creating a new survey.
We'll click on the big green button, give
| | 01:11 |
the survey a name, Hansel&Petal.
Don't worry, though.
| | 01:18 |
This name's only for your reference.
It won't be shown to participants.
| | 01:22 |
Now, let's hit the Let's Go button.
Immediately, you're given a URL for your survey.
| | 01:29 |
Don't worry too much about copying this
down because it will be available to you
| | 01:32 |
throughout the process.
This is the address that you'll give to
| | 01:36 |
participants so they can do the card
sorting exercise.
| | 01:39 |
Underneath the URL, you'll see some tabs.
These are where you'll set up your card sort.
| | 01:46 |
Let's go through each tab in turn.
The Survey tab is where we set the general
| | 01:51 |
information for this sort.
We already gave it a name.
| | 01:55 |
And here we can change the URL the
participants will see.
| | 01:58 |
If you'll be running your sort in a
different language, OptimalSort gives you
| | 02:03 |
the option of showing the instructions and
the rest of the interface in that language too.
| | 02:08 |
We'll stick to English.
Next is an important setting.
| | 02:12 |
How will the survey finish?
You can choose to make it so that you
| | 02:15 |
close it when you're ready, when a certain
date is reached, when a certain number of
| | 02:20 |
people have completed the sort, or both.
I normally only invite a set number of
| | 02:26 |
participants, so I close it when I'm
ready.
| | 02:29 |
The second tab, Cards, is where we'll add
the labels for those cards that will be sorted.
| | 02:36 |
I'm going to use the same list that I had
for my paper sort.
| | 02:38 |
It's saved in an Excel spreadsheet, so
I'll select the data in Excel, copy it,
| | 02:47 |
and then paste it into the field in
OptimalSort.
| | 02:51 |
There's a nice check at the bottom of how
many cards the software thinks I have.
| | 02:55 |
If that's not right, make sure you've
selected them all from Excel and they've
| | 02:59 |
pasted across properly, one on each line.
You have the option to add descriptions,
| | 03:05 |
URLs, or even links to images for each
card.
| | 03:08 |
But we're going to stick to just using
labels.
| | 03:11 |
There are other options on this page, but
we'll stick with the defaults.
| | 03:16 |
We want to randomize the card order, and
we want each participant to see all the cards.
| | 03:21 |
But we won't force them to sort every
card.
| | 03:23 |
That's because forcing people can make
them frustrated, and then they just throw
| | 03:28 |
the cards in any group.
That messes up our sort more than letting
| | 03:31 |
them leave the card unsorted.
The third tab, Categories, is most useful
| | 03:37 |
for closed sorts.
Those are ones where you define the groups
| | 03:41 |
before participants start sorting.
We're using an open sort, where
| | 03:45 |
participants define their own groups.
So we don't have to change anything from
| | 03:48 |
the default settings here.
The fourth tab is important because it's
| | 03:53 |
where you communicate the instructions to
your participants.
| | 03:56 |
Now, if you aren't used to creating
surveys and the language you should use, I
| | 04:00 |
suggest leaving most of this page as it
is.
| | 04:04 |
The Optimal Workshop people spent some
time making sure their instructions work well.
| | 04:08 |
I would, however, change the name from
Optimal Sort to something more relevant,
| | 04:12 |
perhaps Hansel & Petal.
Let's do that now.
| | 04:17 |
Hansel & Petal.
Also, if you're paying your participants,
| | 04:22 |
you might want to mention how they can
claim their gratuity, or how you'll send
| | 04:26 |
it to them, in the Thank you message
further down on the screen.
| | 04:29 |
Let's scroll past the instructions, and
here's the Thank you message.
| | 04:34 |
Maybe we'd say something like, we will
send your amazon.com gift certificate to
| | 04:45 |
the email address you entered earlier.
There.
| | 04:53 |
Now, of course, if we change the thank you
message, we also need to gather the email information.
| | 04:59 |
We do that on the Questionnaire tab.
You'll notice at the top of this section,
| | 05:04 |
there's an area for participant
identification.
| | 05:06 |
By leaving the default, which is Email
address, selected, we'll automatically
| | 05:12 |
collect that information.
You can also see that on this page there
| | 05:15 |
are options to add pre and post survey
questionnaires.
| | 05:21 |
For instance, I might want to know how
experienced each participant was, so I'd
| | 05:25 |
click the Add new post-survey question
button.
| | 05:30 |
And I'd say, how frequently do you
purchase flowers online?
| | 05:37 |
I'll choose a radio button response and
add four options.
| | 05:46 |
My options are going to be Less than once
per year, Once or twice a year, Most
| | 05:56 |
months, and Weekly.
We'll require that they answer the
| | 06:03 |
question, and then finish editing.
You can see how it's possible to ask any
| | 06:08 |
number of questions before or after the
sort takes place.
| | 06:12 |
My suggestion is that you ask as few as
possible, if any, because the card sort
| | 06:18 |
itself will take quite a bit of time.
And you should already have most of the
| | 06:21 |
information you need about participants
from the recruiting process.
| | 06:24 |
If you're using a paid plan, you can also
set options for Access and Appearance.
| | 06:32 |
Those are the final two tabs.
Access allows you to set a password so
| | 06:37 |
that only known participants can get in.
Appearance lets you use your own logo and
| | 06:43 |
color scheme.
That can make your survey seem more
| | 06:46 |
official, and so, more trustable.
After you've made any changes to those
| | 06:52 |
settings, you can hit Save, and your card
sort is ready to go.
| | 06:56 |
It's that simple.
| | 06:58 |
| | Collapse this transcript |
| Running an online OptimalSort card sort| 00:00 |
Now that we've got our card sort set up,
it's almost time to send out the link to
| | 00:04 |
the participants we recruited, and get
them sorting our cards.
| | 00:08 |
There are a couple of things we need to do
first though.
| | 00:11 |
I always like to check the sort is working
how I expected before I send it live.
| | 00:15 |
You can do that using the preview button.
Notice it gets saved, but you see exactly
| | 00:20 |
what your participants would see.
This is your last chance to check that the
| | 00:24 |
card labels are correct because once
assort gets launched, you can't change
| | 00:29 |
those anymore.
If you look to the right of the screen,
| | 00:32 |
you'll notice that optimal sort allows you
to print your cards off so that you can do
| | 00:37 |
what they call a moderated sort.
In other words, an in person card sort.
| | 00:40 |
Then, after you've done the sort, you can
enter the card groupings back in to take
| | 00:45 |
advantage of the analysis tools they
offer.
| | 00:48 |
It's worth bearing this in mind, the
benefits of in-person sorting of physical
| | 00:52 |
cards with the benefits of computer-based
analysis tools.
| | 00:55 |
This option is only available on paid
accounts.
| | 00:58 |
For the moment, we're talking about
performing an on-line sort.
| | 01:03 |
Now it's time to launch your sort.
Just click the launch now button and your live.
| | 01:08 |
Of course your participants still need to
know where to go to complete the sort so
| | 01:14 |
it's time to compose an e-mail to them.
I suggest you think carefully about how to
| | 01:19 |
phrase the e-mail you send to
participants.
| | 01:22 |
There are a couple of important points to
make.
| | 01:24 |
For instance, why are you asking people to
perform the sort for you?
| | 01:29 |
How long it will take.
What they get for doing it?
| | 01:32 |
When you want them to do it by, and who to
contact with questions?
| | 01:37 |
Remember to copy the URL from your
administrative page into the email.
| | 01:42 |
It's probably also worth reminding
participants not to forward the email to
| | 01:46 |
other people, because you don't want your
card sort results with polluted with
| | 01:50 |
people who aren't representative users.
We've put an example email in the
| | 01:54 |
exercises folder of this course for you to
edit and use.
| | 01:58 |
Once you've sent the email, you can track
responses as they come in by looking at
| | 02:03 |
your survey summary page.
On the right hand side is a tally of the
| | 02:07 |
number of responses.
At any time, you can click on the View
| | 02:11 |
Results button and track the answers that
people have given.
| | 02:16 |
At this point, while your survey is still
underway; it's good to keep track of a
| | 02:20 |
couple of statistics just to see whether
things are going to plan.
| | 02:24 |
But first, is the time taken on the
overview tab, if this time looks
| | 02:29 |
suspiciously short, then it's likely the
people aren't giving your sort the full
| | 02:34 |
attention or they're failing to sort all
the available cards.
| | 02:38 |
You can get more detail about this on the
participants tab.
| | 02:43 |
And potentially contact people to appear
had trouble.
| | 02:46 |
Maybe people who only sorted 25% of the
cards, or only created two or three categories.
| | 02:53 |
We'll cover the other tabs in the results
screen in the next section on data analysis.
| | 02:58 |
| | Collapse this transcript |
| Reviewing what participants see| 00:00 |
At this point, I'm going to show you what
a card sort looks like from a participant perspective.
| | 00:05 |
After they click on the link from the
e-mail you send out, this is what
| | 00:09 |
participants see in their browser.
Once they've entered their e-mail address,
| | 00:14 |
they would be shown the presort questions.
We didn't have any presort questions.
| | 00:18 |
So our participants are taken straight to
the sorting interface.
| | 00:22 |
Now you can see why I suggest limiting the
number of cards you ask people to sort.
| | 00:26 |
Although it's possible to scroll the card
sorting area, it's not that large.
| | 00:31 |
Here participants see the first set of
instructions which point out the items
| | 00:35 |
they'll need to sort and tell them how to
proceed.
| | 00:39 |
Let's drag the first item across.
Dragging it across creates a new group
| | 00:45 |
with a space to name the group.
It also replaces the first set of
| | 00:49 |
instructions with some new ones, to link
participants so that they can continue to
| | 00:52 |
drag items across, and to name the groups
whenever they want to.
| | 00:57 |
As participants continue to add more
cards, they can create new groups, or add
| | 01:01 |
to existing ones.
This continues until all the cards are sorted.
| | 01:06 |
Once they've sorted all the cards, and
given each group a name, participants can
| | 01:11 |
finish out the sort as a space to leave
comments if they want.
| | 01:20 |
And the software checks that all the
groups have been named before letting
| | 01:23 |
people continue.
Next, participants see any post sort questions.
| | 01:30 |
This is where we put our radio button
question about purchase frequencies, so it
| | 01:33 |
appears on the screen now.
Let's answer, weekly.
| | 01:39 |
After answering any questions,
participants are shown the final thank you screen.
| | 01:43 |
This is the one we changed to add the
information about gift cards.
| | 01:47 |
At this point, the sort is finished and
the data is captured in the software,
| | 01:50 |
ready for us to analyze.
| | 01:51 |
| | Collapse this transcript |
| Checking your data| 00:00 |
Just like with a paper sort the results
from an online sort can be a bit messy.
| | 00:04 |
Also, because we weren't there to watch
the sort take place we need to check that
| | 00:09 |
people actually performed the sort
properly, rather than just randomly
| | 00:12 |
sliding cards around in order to get their
gratuity.
| | 00:14 |
There are a couple of steps we need to go
through to check that we got valid results.
| | 00:19 |
And to clean up the data before we start
the analysis process.
| | 00:23 |
Let's get back to our card sort result by
clicking on the View Results button for
| | 00:27 |
the Hansel and Petal survey in our list of
surveys.
| | 00:31 |
We already talked about the Overview and
Participants tab, but it's worth pointing
| | 00:36 |
out that you can choose to ignore or even
delete any participant data by choosing
| | 00:41 |
those options for a participant on the
Participants tab.
| | 00:44 |
You might decide to do this if a
participant contacts you and says they had
| | 00:47 |
trouble with the sort then started it
over.
| | 00:49 |
You can delete their first attempt so it
doesn't mess up your data.
| | 00:54 |
This page also lets you do some clever
things, like select a sub group of
| | 00:58 |
participants to analyze based on answers
to questions you asked them.
| | 01:02 |
That's very useful if you want to find out
where the different sub groups of
| | 01:05 |
participants think about the information
architecture in different ways.
| | 01:08 |
For now though, we'll include all of our
participants in the analysis.
| | 01:14 |
Looking at the answers to the questions we
asked, we can see that we get a good cross
| | 01:17 |
section of user types in our participants.
They all had at least one or two purchases
| | 01:22 |
a year.
That's what we wanted.
| | 01:24 |
The less than one per year option was
designed to filter our people we didn't
| | 01:29 |
want, so the results here are reassuring.
The Cards tab lets us know what categories
| | 01:34 |
each card was sorted into.
Cards that are sorted into many more
| | 01:38 |
categories are probably the ones that have
the least agreement between participants,
| | 01:42 |
but before we can be sure we also need to
look on the Categories tab.
| | 01:48 |
This tells us the names that each
participant gave to the groups of cards
| | 01:51 |
they created.
You'll probably find several categories
| | 01:54 |
that mean the same thing.
Like here, we have about listed three
| | 01:57 |
times, and then about us, and about
questions.
| | 02:02 |
Let's look at the cards that were placed
in this last group to see whether it
| | 02:06 |
really means the same thing as the ones
above it.
| | 02:09 |
After clicking on the Plus we can look at
those cards and see that yes indeed, they
| | 02:13 |
are pretty much the same as the ones in
the sections above.
| | 02:16 |
So, we'll select them all, just by
clicking on them, and choose to
| | 02:25 |
standardize the selected categories.
We'll give them a category name and create
| | 02:31 |
the category.
Standardizing our category names makes a
| | 02:35 |
subsequent analysis much easier.
But don't go overboard.
| | 02:39 |
Always click on the Plus under a category
to see the items that people put in it.
| | 02:43 |
Even if a category name sounds similar,
the items people chose may suggest that
| | 02:48 |
the categories should be kept apart.
Now, when we go back to the Cards tab,
| | 02:52 |
you'll see that some of the cards have a
lower number of categories against them.
| | 02:57 |
That's because the standardization has
reduced the number of categories.
| | 03:01 |
Once you've gone through all the category
names and standardized them where it seems
| | 03:04 |
appropriate, you're ready to look at the
Cluster Analysis tools that Optimal Sort provides.
| | 03:09 |
That's what we'll do next.
| | 03:10 |
| | Collapse this transcript |
| Using the built-in analysis tools| 00:00 |
One major advantage of using a computer
based or online tool, is that it can
| | 00:04 |
collect information for you.
So you don't have to do as much data
| | 00:08 |
collection and entry.
Most of the tools also give you an
| | 00:11 |
analysis suite, to help you make sense of
the data.
| | 00:14 |
Here, we'll look at the information you
get from optimal sort.
| | 00:17 |
If you though that the hand based card
sort analysis was painful.
| | 00:21 |
You'll be very relieved to know that
optimal sort and the other tools do a lot
| | 00:25 |
of data entry and formatting for you.
Even if you ran the card sort in person,
| | 00:29 |
optimal sort lets you enter the results,
and lets you do cluster analysis
| | 00:33 |
calculations for you.
Having said that, it's still up to you to
| | 00:37 |
make sense of what the results are telling
you.
| | 00:39 |
That still means diving into the data to
some degree.
| | 00:43 |
Computer based tools aren't any more magic
than an Excel spreadsheet.
| | 00:47 |
They might give you more data
visualizations, but you still need to
| | 00:49 |
spend time working out what the data is
telling you.
| | 00:53 |
In the last video, we started
standardizing our categories.
| | 00:57 |
Now, you can see I've reduced the number
of different groupings by merging ones
| | 01:00 |
that different participants created with
different names.
| | 01:03 |
But which contained very similar items.
Once you have your categories standardized
| | 01:08 |
you can go ahead and start your analysis.
Cluster analysis allows you to see where
| | 01:12 |
your participants agreed with each other
about the placement of items.
| | 01:15 |
It builds groups by looking at the
distance between each pair of cards.
| | 01:20 |
Distance is a strange word here.
You can look at it as meaning that cards
| | 01:24 |
that were grouped together more frequently
are closer than cards that are grouped
| | 01:28 |
together less frequently.
Let's start on the Dendrograms tab.
| | 01:32 |
Dendrograms are hierarchical maps, showing
the relationship between items in the card sort.
| | 01:38 |
They don't provide a one on one mapping
with your potential menu structure, but
| | 01:41 |
they are a helpful way of seeing how users
think.
| | 01:44 |
The Dendrograms tab in optimal sort shows
you two different dendrograms.
| | 01:48 |
Pactual agreement on the top and best
merge beneath it.
| | 01:52 |
Both are useful depending upon the type of
card sort you ran.
| | 01:55 |
For us, with a smaller sample, the best
merge method is likely to be a better
| | 02:00 |
representation of our participant's
groupings.
| | 02:03 |
Let's look at what this dendrogram is
telling us.
| | 02:06 |
The shorter the path between any 2 items,
the more similarity those items have with
| | 02:10 |
each other.
So make a custom arrangement and choose
| | 02:14 |
your own flowers for a display, are more
similar in the sort results than are set
| | 02:19 |
up a consultation for a bridal bouquet and
get Valentine's day roses.
| | 02:23 |
Notice also the percentage numbers along
the top of the dandrogram.
| | 02:28 |
Those are a measure of the amount of
agreement.
| | 02:31 |
So there's more agreement between make a
customer arrangement and choose your own
| | 02:35 |
flowers for a display at 93%, than there
is between set up a consultation for a
| | 02:40 |
bridal bouquet and get Valentine's day
roses at 54%.
| | 02:45 |
Don't get hung up on the actual value.
What's important is the relative value
| | 02:49 |
between two pairs of items.
The optimal sort similarity matrix gives
| | 02:54 |
you an indication of how frequently any
two items were paired together.
| | 02:57 |
The color coding allows you to see the
strength of the pairings between any two cards.
| | 03:02 |
Using our two examples from the
dendrogram, you can see that make a custom
| | 03:06 |
arrangement and choose your own flowers
for a display have a higher similarity
| | 03:11 |
with 14 agreements.
Than do setup a consultation for a bridal
| | 03:16 |
bouquet, and get Valentine's day roses,
with seven.
| | 03:20 |
That suggests that the first pairing is
more similar in the sort results.
| | 03:25 |
Now let's move onto the standardization
grid tab.
| | 03:29 |
The standardization grid is a tool that we
talked about in chapter four.
| | 03:33 |
This is just a list of the cards that
people placed in each of the standard
| | 03:36 |
group names that you created by merging
participants' category names.
| | 03:40 |
It tells you how often each card was
placed within each of the standard group names.
| | 03:45 |
Now, obviously, you still have to think of
standardized names yourself, but
| | 03:49 |
OptimalSort creates the grid for you.
Something different about optimal sort is
| | 03:54 |
the participant-centric analysis on the
PCA tab.
| | 03:56 |
This shows the top three sorts as ranked
by the number of other sorts that used a
| | 04:03 |
similar type of grouping.
In other words, the groupings shown here
| | 04:07 |
indicate the most popular or most
representative groupings.
| | 04:10 |
This can be very helpful in deciding what
your final information architecture should
| | 04:14 |
look like, because it lets you see the one
set of groupings that most people sorts
| | 04:18 |
were similar to.
Obviously, if you had different types of
| | 04:22 |
participant complete the sort, you might
expect the participant-centric analysis to
| | 04:27 |
show you a representative sort from each
of those participant types, if they truly
| | 04:31 |
do think about the information
architecture in different ways.
| | 04:34 |
One other thing to mention here, is how
easy OptimalSort makes it to export your
| | 04:40 |
data to use in your own analysis.
The downloads tab gives you access to both
| | 04:45 |
raw data files and to standardized data,
if you study grouping categories.
| | 04:49 |
This means you can use OptimalSort just as
a data collection tool, as a partial
| | 04:55 |
analysis tool, or you can use it to build
an interpretations for all your analysis.
| | 04:59 |
Donna Spencer is one of the key people in
the information architecture field.
| | 05:04 |
She's created this spreadsheet that helps
with analysis.
| | 05:07 |
If you're familiar with her work, you'll
be happy to see that OptimalSort creates
| | 05:10 |
output in that format.
I'll give you more information on Donna's
| | 05:14 |
work at the end of this course.
Now that you've got all your data
| | 05:17 |
standardized and created your cluster
analysis results, you can apply the same
| | 05:21 |
techniques that we talked about in chapter
four to eyeball your data.
| | 05:24 |
The cluster analysis output gives you a
couple more perspectives, but it's still
| | 05:28 |
up to you to draw together all the threads
and apply some creativity to come up with
| | 05:32 |
your information architecture.
| | 05:33 |
| | Collapse this transcript |
|
|
6. Creating an Information Architecture from Your AnalysisStarting with an abstract structure| 00:00 |
Your ultimate aim is to have a menu
structure and also the other navigation
| | 00:04 |
elements for your site.
You might also want to know how to
| | 00:08 |
categorize content on the site.
But we aren't quite there yet.
| | 00:12 |
First, we have to create out abstract
information architecture.
| | 00:15 |
This provides us with a description of how
users think about and classify the stuff
| | 00:20 |
that's on our site, without any of the
constraints imposed by the site's design.
| | 00:24 |
It's important to have this overview
description.
| | 00:28 |
Because if we jump straight into designing
menus we may miss the big picture.
| | 00:32 |
Some parts of your information
architecture won't be displayed in menus,
| | 00:37 |
and bits which are could be shown in
different places.
| | 00:40 |
For instance, you might make a distinction
between site tools and site content.
| | 00:45 |
Displaying each in it's own menu, or you
might decide that news and events should
| | 00:51 |
perform the basis of the site's homepage
and thus potentially not need a main menu item.
| | 00:57 |
Similarly, support areas could either be
displayed as a menu item or as links in
| | 01:02 |
the page footer.
So, in order to make sure we put the right
| | 01:05 |
things in our menus, but also don't forget
any important groupings that users consider.
| | 01:09 |
We need to first create our abstract
information architecture before we apply
| | 01:14 |
it to our site design.
Cluster analysis and dendrograms show us
| | 01:20 |
how participants grouped the card sort
items.
| | 01:23 |
The comments that participants made during
the card sort sessions can help you
| | 01:26 |
interpret these groupings.
It's important to point out that none of
| | 01:30 |
the data visualizations that the software
analysis tools provide will tell you
| | 01:34 |
directly how to build your website menu
structure.
| | 01:37 |
For instance, although the dendrogram
output looks a bit like a hierarchical
| | 01:42 |
split of the information, you have other
decisions to make before creating the
| | 01:46 |
actual menu structure.
The best thing to do at this point, is to
| | 01:49 |
create the abstract information
architecture, showing the ideal structure
| | 01:53 |
of the information regardless of how it
will be used.
| | 01:56 |
Then, you can apply some real world
constraints to that structure, so that it
| | 02:01 |
informs and fits in with your real world
site design.
| | 02:04 |
| | Collapse this transcript |
| Creating abstract information architecture| 00:00 |
Before we roll our sleeves up and design
our site's menus and other components, we
| | 00:04 |
need the abstract information
architecture.
| | 00:06 |
To make this abstract information
architecture, we take all of our
| | 00:11 |
observations, along with a cluster
analysis information and create our best
| | 00:15 |
impression of the ideal structure for that
information, regardless of how it will be used.
| | 00:21 |
Items that participants grouped most
clearly will form the basic lower level structure.
| | 00:25 |
For instance, one grouping consisted of
minimal impact flower borders, native
| | 00:30 |
landscaping, yard plants, and plants that
would require no extra water.
| | 00:37 |
We'll form several search groupings.
Then, the lower level groupings that have
| | 00:41 |
the most in common, in participants minds,
form the categories above this lower level structure.
| | 00:47 |
Names that participants gave to the
groupings might suggest category headings.
| | 00:52 |
You probably won't choose to use the exact
names that participants wrote down, but
| | 00:56 |
their terminology and the thinking behind
their choices will inform your final decisions.
| | 01:02 |
If you created standardized names, those
might be closer to the final names you'll
| | 01:06 |
plan on using.
Names that people gave to this group were
| | 01:09 |
outdoor plants or live plants, but all of
these items describe ecologically sound
| | 01:15 |
outdoor plants which is probably a more
descriptive category.
| | 01:18 |
Your abstract information architecture
hierarchy should emerge from this data.
| | 01:25 |
I have to say there's as much art to this
as there is science.
| | 01:28 |
Luckily, because we'll be verifying our
assumptions by running a reserve card sort
| | 01:32 |
later, it's okay that this isn't the most
rigorous process.
| | 01:37 |
So long as you make sure you're always
working form the data that you have from
| | 01:39 |
your analysis tools or spreadsheet you
should be fine.
| | 01:43 |
A good information architecture structure
will provide a balance between breadth and depth.
| | 01:50 |
Too many top level categories might make
it hard for customers to pick the right option.
| | 01:55 |
Too few might require them to drill down
through too many levels.
| | 01:59 |
It will also provide good sign posts so
that people know where they are at each
| | 02:03 |
point in the architecture based on the
category and sub-category labels.
| | 02:08 |
The right information architecture also
means showing items in the correct order.
| | 02:12 |
That means making sure that sub-category
items are arranged in the order that
| | 02:17 |
people expect or that information can be
sorted and filtered based on a set of
| | 02:22 |
criteria you've derived from user
research.
| | 02:25 |
The arrangement will depend upon the type
of data being displayed and how your users
| | 02:29 |
think about it.
There are several options.
| | 02:32 |
Importance based puts the most important
items first.
| | 02:36 |
Frequency or recency based are the most
common or recent items.
| | 02:41 |
Topic based is used to describe content.
Task based to show available actions.
| | 02:47 |
And time based, depending upon when an
event occurred, for instance, in your web history.
| | 02:53 |
Product based arrangements are split by
feature set.
| | 02:57 |
This is a special type of topic based
classification.
| | 03:00 |
Location based could be geographic
location, by stockroom shelf, or by office number.
| | 03:07 |
Audience based could be categorized by
employee information or customer information.
| | 03:12 |
Series based are things such as shoe or
clothing sizes.
| | 03:17 |
And alpha, or numerical, are based on the
quality of the label of the item, like,
| | 03:21 |
book authors.
Across your site you might use several
| | 03:26 |
different orders.
The order of top level menus might be
| | 03:29 |
frequency based so that the most common
appear first or they might be a built-in
| | 03:35 |
arrangement dictated by the type of task
being performed on the site.
| | 03:40 |
Lower level menus could be sorted pretty
much any way.
| | 03:42 |
Again, depending upon the type of data
they contain.
| | 03:46 |
Homepage content might show the most
popular items, which is based on frequency
| | 03:51 |
of purchase.
Often these days, footer areas contain the
| | 03:54 |
list of product categories and on
corporate sites, they might also have
| | 03:58 |
content for different audiences, such as
the press, investors, and job seekers.
| | 04:03 |
So you can see how your categorizatoin
decisions are driven not just by the
| | 04:08 |
groupings the particpants choose, but also
the types of use that you expect them to
| | 04:12 |
have for the items in those gropuings.
| | 04:15 |
| | Collapse this transcript |
| Knowing the problems you might face| 00:00 |
As you go through the results, you'll see
some clusters where there is obvious
| | 00:04 |
agreement among participants, and others
where there is less agreement.
| | 00:08 |
The items that have little agreement will
appear in different places for each
| | 00:11 |
participant, or sometimes participants
might make a pretty obvious other or
| | 00:16 |
miscellaneous cluster.
This might indicate that participants
| | 00:20 |
didn't understand the item, that they
don't really care about it, or that it
| | 00:25 |
didn't fit with the rest of the site's
structure and content.
| | 00:29 |
You'll have to work out which of these
explanations is most likely, and then do
| | 00:33 |
something about it.
It could be that the terminology you use
| | 00:36 |
on the site needs to change to be more in
line with your customers vocabulary.
| | 00:41 |
It might be that you've recruited
participants who weren't representative
| | 00:44 |
users for that part of the site.
Or it could really be that your site is
| | 00:48 |
trying to merge concepts that just don't
go together in people's minds.
| | 00:53 |
The comments the participants made during
the card sort session should be helpful to
| | 00:57 |
you here.
Whatever you decide is the answer, you
| | 01:00 |
need to do some more research work to
verify your assumptions and then correct them.
| | 01:05 |
That might involve finding more
participants of the type that would use
| | 01:08 |
certain areas of the site.
Or it might be a deeper issue to do with
| | 01:12 |
how you name your products or services.
And until you fix that issue, you just
| | 01:16 |
won't be able to create a suitable
information architecture.
| | 01:20 |
Another problem that people sometimes have
with card sort data is that there's not a
| | 01:24 |
single, clear hierarchy.
Sometimes the cluster analysis doesn't
| | 01:28 |
show clear areas of agreement, or the
dendrogram starts linking groups of items
| | 01:33 |
that don't logically belong together.
If you have trouble finding the hierarchy,
| | 01:37 |
it might be worth trying to create it two
different ways.
| | 01:41 |
In the past, I've noticed that different
groups of users sometimes think about
| | 01:45 |
content or tasks in different ways.
Using our Hansel and Petal site as an
| | 01:50 |
example, it might be that users who are
interested mainly in buying cup flower
| | 01:54 |
bouquets as gifts see the site organized
one way, while other users who are more
| | 01:59 |
interested in horticulture think about the
site organization a completely different way.
| | 02:03 |
The best way to create two different
hierarchies is to split your participants
| | 02:09 |
into two different groups and perform the
cluster analysis separately for each group.
| | 02:14 |
To do this, you have to know enough about
your participants to be able to split them out.
| | 02:20 |
Let's say you decided, based on
demographic information and their comments
| | 02:24 |
during the session, that ten of your
participants were bouquet buyers, and five
| | 02:28 |
were serious gardeners.
By running the cluster analysis separately
| | 02:33 |
on the results of these two groups, you
should end up with more agreement within
| | 02:37 |
each group, even if overall you now have
two separate hierarchies.
| | 02:42 |
Creating two hierarchies isn't such a big
problem.
| | 02:45 |
It's often possible to design navigation
systems that incorporate the two ways of
| | 02:49 |
thinking and to use one set of concepts in
one area of the site and a different set
| | 02:54 |
of concepts elsewhere.
It's better to find this out now and deal
| | 02:59 |
with it, rather than struggling to
retrofit it after you've created your site
| | 03:03 |
or application.
| | 03:04 |
| | Collapse this transcript |
| Understanding that card sorting isn't a precise technique| 00:00 |
As you've probably realized, card sorting
is not a precise technique.
| | 00:05 |
Different participants will sort cards in
different ways, and so the cluster
| | 00:08 |
analysis and dendrogram won't magically
produce the ideal information architecture
| | 00:13 |
for you.
However, once you start diving into your
| | 00:17 |
own data, you'll quickly start seeing the
patterns and the strength of the relationships.
| | 00:22 |
You'll also be able to cross reference
that, with what you heard individual
| | 00:26 |
participants say as they performed the
sort.
| | 00:29 |
Working from the Cluster analysis, the
Dendrogram, the raw sort results and the
| | 00:33 |
notes you took of participants' comments,
you can pull together a suitable grouping
| | 00:37 |
of items.
Of course, there might also be business
| | 00:41 |
rules or real world contraints that mean
certain items have to go in certain places.
| | 00:47 |
Sometimes, unfortunately, politics gets in
the way of a good information architecture.
| | 00:52 |
Luckily, you can also use the data from
your sessions to help convince management
| | 00:56 |
that it's time for a change.
I like to print all the information off,
| | 01:02 |
arrange it on my desk, and just absorb it
for awhile.
| | 01:04 |
Then, I try creating groups that seem to
best match the majority view based on the
| | 01:10 |
sort results.
I'll check the groups I create against the
| | 01:13 |
individual raw data, in case there was
some people who sorted an entirely
| | 01:16 |
different way.
But my aim is to make a hierarchy that
| | 01:19 |
will be acceptable to everyone who
participates in the sort.
| | 01:22 |
In the end, you need to apply a
combination of knowledge from the cluster
| | 01:27 |
analysis, dendrogram, and what
participants said during the sort, in
| | 01:31 |
order to create a good first pass at an
information architecture.
| | 01:35 |
You should also use other data you have,
such as usability studies, customer
| | 01:39 |
support data, search logs, and web logs,
to inform your analysis.
| | 01:44 |
Don't blindly follow the statistical
output.
| | 01:47 |
Think about what participants said, and
about real world implications.
| | 01:51 |
Back up your decisions with this data.
If you can't find data to back up your
| | 01:56 |
decision, it indicates you might not have
it right.
| | 02:00 |
The next step, after we have our abstract
hierarchy, is to refine it by testing it
| | 02:05 |
with a reverse sort.
That's the next topic.
| | 02:07 |
| | Collapse this transcript |
|
|
7. Validating the Information Architecture with Reverse SortingMaking sure your hierarchical structure is correct| 00:00 |
Once you've put together a draft of your
structure, you should test it with users
| | 00:04 |
by doing a reverse card sort.
This allows you to get quick feedback that
| | 00:08 |
the assumptions you made when you created
your new information architecture are
| | 00:12 |
right, before you make any design changes.
A reverse card sort is very much like it sounds.
| | 00:20 |
Rather than sorting the task cards into
groups, participants indicate where they
| | 00:24 |
think they'd go within your information
architecture hierarchy to perform each
| | 00:28 |
task or find each item.
We still run the reverse sort with our
| | 00:33 |
abstract information architecture
hierarchy, rather than the menu structure
| | 00:37 |
we think we'll ultimately create.
That way, we get to test out the whole
| | 00:41 |
structure in one go.
The reverse sort, which is sometimes also
| | 00:46 |
called tree testing because your hierarchy
is a bit like a tree structure, gives you
| | 00:51 |
a good sense as to whether users can or
can't find items.
| | 00:56 |
You can count the number the hits, that is
times when participants placement of items
| | 01:00 |
matched your expectations, and turn this
into a percentage score that shows the
| | 01:05 |
agreement between your information
architecture and participants' ideas of
| | 01:09 |
what's right.
Running the reverse sort face to face with
| | 01:13 |
your participants also means you can hear
their reasons for choosing certain parts
| | 01:17 |
of the hierarchy, which helps you work out
the reasons why some parts of the
| | 01:20 |
hierarchy aren't working as well as you'd
like.
| | 01:24 |
As with card sorting, there are both
paper-based and computer-based ways of
| | 01:28 |
performing reverse sort.
I like the immediacy of a paper-based
| | 01:31 |
reverse sort.
But sometimes you might need to reach a
| | 01:34 |
wider audience or let participants
complete the sort in their own time.
| | 01:38 |
So the online tools can be a great
benefit.
| | 01:41 |
We have both types of reverse sort in this
chapter, starting with the paper-based sort.
| | 01:46 |
| | Collapse this transcript |
| Creating and running a paper-based reverse sort| 00:00 |
Running a paper-based reverse sort is
really simple.
| | 00:03 |
And the analysis is much easier than for a
card sort because you just have to create
| | 00:07 |
a tally of where each participant placed
their task cards.
| | 00:12 |
The output is just a count of the number
of places where participants agreed and
| | 00:16 |
disagreed with your hierarchy, plus any
comments they made along the way.
| | 00:21 |
The way we run the reverse sort is to
create an index card for each part of the
| | 00:25 |
information architecture hierarchy that we
created.
| | 00:28 |
Then we get people to tell us where in
this hierarchy they'd expect to find
| | 00:32 |
certain items.
These items correspond to the cards that
| | 00:36 |
we used for our initial card sort.
If participants expectations are a good
| | 00:41 |
match with the hierarchical structure, we
know we're on the right track.
| | 00:46 |
We would typically use between 15 to 20
participants again, if they're all
| | 00:51 |
representative of the same user type.
We can follow the same basic protocol as
| | 00:55 |
with the original card sort.
But because the reverse sort uses a
| | 00:59 |
slightly different method, I've included a
different moderator script for you to use.
| | 01:04 |
You can find it in the exercise files for
this course.
| | 01:08 |
To run the reverse sort, you'll need two
sets of cards, the navigation deck with
| | 01:14 |
the information architecture hierarchy and
the task deck, which is the cards from the
| | 01:18 |
initial card sort.
I also make four cards for participants to
| | 01:23 |
place the task cards on when they're done
with them.
| | 01:26 |
I use labels of Found it, Confusion,
Terminology, and Give up.
| | 01:33 |
You can of course, make whatever changes
you want.
| | 01:36 |
These labels seem to generate the most
conversation, however.
| | 01:39 |
Participants have the whole deck of task
cards and take each one in turn.
| | 01:44 |
They read the task out loud and then tell
you which navigation choice they would
| | 01:49 |
expect to find it under.
You don't show all of the navigation deck
| | 01:53 |
cards at once.
For each task, start with just the top
| | 01:57 |
level card showing.
Participants will tell you which menu item
| | 02:01 |
they would choose, and then you place the
index card associated with that item on
| | 02:06 |
the table.
If you've reached the lowest level in your
| | 02:09 |
hierarchy, you just tell them so.
At this point, the participant should tell
| | 02:13 |
you whether this is where they expected to
find the item.
| | 02:17 |
If it is, you tally the result and move
on.
| | 02:20 |
If it isn't, let the participant direct
you back up through the structure and then
| | 02:26 |
down to an alternative location in the
hierarchy.
| | 02:31 |
Only let the participant go down one
alternate location before moving on to the
| | 02:35 |
next card.
Because in real life, it's unlikely that
| | 02:39 |
people would have much more patience for
your site if they couldn't find the
| | 02:42 |
information after two separate tries.
After each task, tally the results on a
| | 02:48 |
piece of paper.
Write down the task number, and then the
| | 02:51 |
path the participant took, and whether
they placed the card on the Found,
| | 02:56 |
Confusion, Terminology, or Give up group.
Also note down any comments the
| | 03:02 |
participant made during the session.
| | 03:04 |
| | Collapse this transcript |
| Analyzing a paper-based reverse sort| 00:00 |
The results from a reversed sort are easy
to visualize in a simple Excel table.
| | 00:05 |
Just tally the number of participants who
chose a particular menu location for each task.
| | 00:11 |
Here, you can see menu locations on the
left hand side and tasks across the top.
| | 00:17 |
You should highlight the cells in the
spreadsheet that correspond with the
| | 00:20 |
correct answer.
In other words, where you would expect
| | 00:24 |
people to go to find the information.
Here we've highlighted them in green.
| | 00:29 |
For some tasks, there may be more than one
correct location.
| | 00:33 |
Once you've added all the data, you'll
find that most cells in the table will
| | 00:37 |
remain blank.
If your information architecture was spot
| | 00:41 |
on, all the participants will have chosen
the locations that you wanted them to and
| | 00:45 |
so you'll have a large tally against one
navigation menu item for each task.
| | 00:50 |
For some tasks you'll see a distribution
of responses between a couple of areas in
| | 00:54 |
the navigation menu structure.
That indicates that you either need to
| | 00:58 |
improve the differentiation between the
areas, or provide ways for users to
| | 01:02 |
complete that task from either location.
Either by adding a related link style link
| | 01:08 |
or by duplicating content.
If you see a large spread of responses for
| | 01:12 |
a task Though either of the task was
ambiguous and you'll pick this up from
| | 01:16 |
whether it was placed on confusion or
terminology group, or the menu stretcher
| | 01:21 |
isn't supporting that task for your user
group.
| | 01:24 |
Look at the comments that participants
made for that task to get more insight.
| | 01:28 |
You can also count the successful hits and
turn that into a percentage comparison
| | 01:34 |
score for against subsequent iterations of
the information architecture structure.
| | 01:38 |
For instance, if ten out of 15
participants chose the correct area,
| | 01:42 |
that's a 66% hit rate.
If 12 out of 15 chose it, that's an 80%
| | 01:47 |
success rate.
| | 01:49 |
| | Collapse this transcript |
| Interpreting the results| 00:00 |
The main piece of information that you get
from a reverse sort is how much agreement
| | 00:05 |
there was between your information
architecture hierarchy and your
| | 00:08 |
participant's expectations.
That agreement metric is the proportion of
| | 00:13 |
times that people managed to find the
correct location for an item from all the
| | 00:17 |
different attempts.
If you want, you can also track the
| | 00:21 |
proportion you found it on first, as
opposed to their second attempt, which is
| | 00:25 |
a measure of directness to the answer.
Only you can decide what level of
| | 00:31 |
disagreement you're prepared to accept
between your hierarchy and your
| | 00:35 |
participants' expectations.
You're highly unlikely to get 100%
| | 00:39 |
agreement, and that's okay.
Participants will be much more likely to
| | 00:43 |
find items in the real world because
they'd have many more visual cues such as
| | 00:48 |
images and content text to let them know
whether they're on the right track or not.
| | 00:53 |
Directness tells you how confident people
were that they'd find an item.
| | 00:58 |
If certain items have a low directness.
In other words, they were found more on
| | 01:02 |
the second than on the first attempts,
then that indicates that the structure was
| | 01:05 |
working, but not as well as it could.
In the past, I've run studies where
| | 01:12 |
agreement went from 35% with the existing
system, to 75% in our first round of
| | 01:17 |
reverse sorting.
Rising to 90% after some subsequent tweaks.
| | 01:22 |
It's always interesting to run a reverse
sort using the existing architecture,
| | 01:27 |
because that way, you'll have a great
indication of how much your changes are
| | 01:31 |
likely to improve the system.
You can be sure that the 35% figure was a
| | 01:36 |
shock to the team who's navigation
structure we tested.
| | 01:39 |
If the level of disagreement is too high
after you reverse sort, or if your
| | 01:43 |
directness measure is too low for some
items, you can use the results to see
| | 01:47 |
where participants would have expected the
item to be placed.
| | 01:51 |
That will help you make changes.
Although remember that just moving items
| | 01:55 |
between branches of the heirarchy might
not be the answer.
| | 01:59 |
Your hierarchy could be missing a crucial
category.
| | 02:02 |
The only way to find that out is by
analyzing what participants said as they
| | 02:06 |
completed the reverse sort.
Obviously that's harder if you did your
| | 02:10 |
reverse sort online although some of the
online tools do give your participants an
| | 02:14 |
opportunity to type in comments on the
process.
| | 02:17 |
Once you're happy with a level of
agrteement, you have an information
| | 02:20 |
architrecture that you can use for the
next step, which is building your site's layout.
| | 02:25 |
| | Collapse this transcript |
|
|
8. Computer-Based Reverse SortingExploring computer-based reverse sorting| 00:00 |
As with card sorting, there are software
tools to help you present the reverse
| | 00:04 |
sorting cards.
Most of these tools also collate and
| | 00:08 |
present the results for you in ways that
can help you interpret the data.
| | 00:12 |
The tools normally work by displaying the
top level items in your information
| | 00:15 |
architecture as a menu.
Participants are given tasks and asked
| | 00:20 |
where they'd expect to find the answer.
When participants select a certain item in
| | 00:24 |
the menu, the tool displays a sub-menu of
the items in the next level of the hierarchy.
| | 00:29 |
This way it's possible to track where each
participant says they would expect to find
| | 00:33 |
the items you specify.
The two main ways that people use computer
| | 00:38 |
based reverse sorting are to automate a
paper card sort session, or to make it
| | 00:42 |
possible to run online reverse sorts.
As with online card sorts, there are
| | 00:47 |
several benefits to conducting a reverse
sort online.
| | 00:49 |
It might give you access to more of your
users, and some tools allow participants
| | 00:54 |
to complete the sort in their own time.
You send a link in an email to each
| | 00:58 |
participant, and then the tool notifies
you when your participants have completed
| | 01:01 |
the sort.
The downside is that you can't watch each
| | 01:04 |
session as it occurs and hear what
participants say.
| | 01:08 |
So the level of feedback you get is lower
than with in-person sorts.
| | 01:13 |
If you don't have time to run sessions or
if your participants are on different time
| | 01:18 |
zones, you might be better off using a
tool that automates the whole process for you.
| | 01:22 |
Some tools you might want to consider for
online reverse sorts are C-inspector,
| | 01:26 |
which is $99 per study, and the tree
testing part of UserZoom suite,which we
| | 01:31 |
mentioned earlier.
Optimal Workshop, who make OptimalSort,
| | 01:36 |
also make a reverse sorting tool called
Treejack, which is a $109 per month for
| | 01:41 |
unlimited use.
That's the product I'm going to talk about
| | 01:44 |
here, to demonstrate how this type of tool
works.
| | 01:47 |
As with card sorting, you'll find that
most of the computer based tools follow
| | 01:51 |
the same process, although some have more
analysis options than others.
| | 01:55 |
| | Collapse this transcript |
| Using Treejack for reverse sorting| 00:00 |
Let's go through the process of setting up
a reverse sort in Treejack, the online
| | 00:04 |
product from Optimal Workshop.
You'll find that the principals are
| | 00:08 |
similar for most other reverse sort
software, even if the interface or
| | 00:12 |
terminology is different.
First, log into your optimal workshop
| | 00:16 |
account, then choose the Treejack tab.
Notice that as a free account user, I can
| | 00:21 |
only run reverse sort with three tasks.
That's not very many.
| | 00:25 |
I want to be running more than that from
my Hansel & Petal survey.
| | 00:29 |
So I'll need to upgrade to a paid account.
I'm going to do that now.
| | 00:34 |
Now I'm back after upgrading, and you'll
notice that the nag screen has disappeared.
| | 00:39 |
It's time to click the Create a New Survey
button and get started.
| | 00:43 |
I'll name my new survey Hansel & Petal,
and then hit the Let's go button.
| | 00:50 |
There is nothing on the first screen that
I need to change.
| | 00:54 |
The default settings are all just fine.
You'll notice that just like in the card
| | 01:00 |
sort set up, I can chose whether to cut
off participation at a certain date or
| | 01:05 |
level of responses, but for this study I'm
happy to close the survey manually when
| | 01:09 |
I'm done.
It doesn't affect analysis at all.
| | 01:12 |
Now, we can add our information
architecture structure into the tool on
| | 01:16 |
the second tab called tree.
We can either type it in manually, or
| | 01:21 |
Paste it in from a spreadsheet.
If your entering the data manually, make
| | 01:26 |
sure there's only one item per line, and
that you indent child items using either
| | 01:30 |
tabs or spaces, but not both.
I use tabs, because it's easier to see
| | 01:34 |
whether I got my indented items correct.
However, today I'm going to Paste from a spreadsheet.
| | 01:41 |
Here you can see my spreadsheet, set out
with child tasks and columns further to
| | 01:45 |
the right.
So the B column is my home column, then C
| | 01:50 |
is my top level categories, and D is the
items under each top level category.
| | 01:54 |
I'll select all my data and then Copy it.
(SOUND) Now I'll Paste my data into the
| | 02:02 |
field in the Tree tab of Treejack.
Clicking the Build at the tree button,
| | 02:08 |
lets me check that the stretch will come
out how I expected.
| | 02:10 |
This is important, because if we make
changes later, it will mean rework in
| | 02:17 |
other areas of the survey too.
Next, we must enter the tasks that we want
| | 02:21 |
participants to perform on the tasks tab.
These are the items that would be on our
| | 02:27 |
task cards, if we were running a paper
based reverse sort.
| | 02:30 |
They're the same items that would have
been on the original card sort cards as well.
| | 02:35 |
Unfortunately, we can't just Copy and
Paste them in as one long list, but
| | 02:38 |
there's a good reason for that.
Each time we add a task, we have to say
| | 02:42 |
where in the tree it should live.
Let me demonstrate by adding a task now.
| | 02:47 |
I have my list of tasks in the
spreadsheet, so I'll go there now.
| | 02:50 |
I'll Copy the first task.
And then go back to Treejack, and Paste it
| | 02:56 |
into the text box.
First I click, Add a new task.
| | 03:01 |
Then I Paste it in.
Now, I have to click the Select correct
| | 03:05 |
answers link.
My tree opens up, and I can select the
| | 03:10 |
correct location for the answer.
I can select more than one location, if
| | 03:14 |
it's possible for the task to be completed
in several places.
| | 03:18 |
Calling the store, what's their number.
That would be under customer service, and
| | 03:23 |
About Us might also be under customer
service itself.
| | 03:28 |
So I'll check both of those boxes, and
click Update correct answers.
| | 03:33 |
Now I'll hit Save changes.
I have to repeat this for every task.
| | 03:39 |
It takes some time, but one benefit is
that it makes you think about whether your
| | 03:43 |
architecture is correct, as you go through
the process of telling the tool where each
| | 03:47 |
answer should be.
When we're done, we'll leave the other
| | 03:50 |
options for skipping tasks and random
presentation, checked, and Save our work.
| | 03:57 |
The next tab, Messages, allows us to edit
the information that participants are
| | 04:02 |
given before and after the study.
I think the optimal work shop have done a
| | 04:07 |
good job with the default text.
You might want to change a couple of
| | 04:10 |
items, to make it sound more like your
companies brand or voice.
| | 04:14 |
Again, as with the card sort, if you're
paying your participants you might want to
| | 04:19 |
mention how they can claim the gratuity,
or how you'll send it to them.
| | 04:23 |
In thank you message, scroll down to that
now.
| | 04:27 |
Maybe something like, we'll send you an
amazon gift certificate to the e-mail
| | 04:31 |
address you entered earlier.
You might want to collect answers to some
| | 04:35 |
demographic questions or other data.
Treejack lets you add question either
| | 04:40 |
before or after the reverse sort is run.
The interface is the same as for the
| | 04:44 |
optimal sort card sort questions.
For the reverse sort, I'm not going to ask
| | 04:49 |
any demographic questions.
If you're using a paid plan, you can also
| | 04:53 |
set options for Access and for Appearance.
Access allows you to set a password so
| | 05:00 |
only known participants can get in.
Appearance lets you use your own logo and
| | 05:04 |
color scheme.
At this point, you're done with the set up
| | 05:08 |
and it's almost time to take your reverse
sort live.
| | 05:10 |
| | Collapse this transcript |
| Running an online reverse sort with Treejack| 00:00 |
>> Now that we've got our reverse source
set up, there're only a couple of things
| | 00:04 |
we need to do, in order to kick it off.
First, let's set the tree coverage graph.
| | 00:10 |
With all my tasks entered already, I click
on the tree coverage link in the task tab.
| | 00:16 |
This shows me a picture of all the nodes
in my information architecture structure
| | 00:20 |
and let me see which one have potential
correct answer on them.
| | 00:24 |
I'm happy that the ones that don't, which
are privacy policy, terms of use, and then
| | 00:29 |
a couple of the special events nodes,
don't need testing.
| | 00:33 |
So my coverage is good.
In other words, participants have the
| | 00:36 |
option to select almost all of my
structure in their answers.
| | 00:40 |
I always like to check that the reverse
sort is working how expected before I send
| | 00:44 |
it live.
You can do this using the preview button.
| | 00:48 |
In order to get save, you see exactly what
your participants would see.
| | 00:52 |
This is your last chance to check that the
hierarchy is correct, because once a
| | 00:55 |
reverse sort gets launched, you can't make
any changes to it.
| | 00:58 |
Now, it's time to launch your reverse
sort.
| | 01:02 |
Just click the Launch Now button and
you're live.
| | 01:04 |
Of course, just like with the card sort,
your participants still need to know where
| | 01:09 |
to go to complete the reverse sort.
So it's time to compose an e-mail to them.
| | 01:13 |
Any time you're sending e-mail to
participants, you need to think carefully
| | 01:17 |
about whether they'll interpret the
instructions the same way you intended.
| | 01:20 |
You need to make the same important points
as in the card sort email earlier in the course.
| | 01:25 |
Why you're asking it to perform the sort
for you.
| | 01:28 |
How long it should take.
What they'll get for doing it.
| | 01:31 |
When you want them to do it by and who to
contact with questions.
| | 01:35 |
Also remember to copy the URL from your
administrative page into the email.
| | 01:39 |
Don't bother trying to give people
instructions on how to complete the
| | 01:43 |
reverse sort in this email.
By the time they get to the site, they'll
| | 01:46 |
have forgotten what you told them.
Instead, rely on the site to give
| | 01:50 |
instructions and context.
We've put an example email in the
| | 01:53 |
exercises folder of this course for you to
edit and use.
| | 01:56 |
Once you've sent the email, you can track
responses as they come in by looking at
| | 02:01 |
your survey administration page.
On the right-hand side of the screen is a
| | 02:05 |
tally of the number of responses.
At any time, you can click on the button
| | 02:09 |
right next to it, and track the answers
that people have given.
| | 02:14 |
You can track that things are running
smoothly by checking a couple of statistics.
| | 02:18 |
We get more information on our reverse
sort of U tab than we did for the card sort.
| | 02:22 |
because there are more direct metrics that
the software can capture.
| | 02:26 |
If you look at the Overview tab in the
Study Results area, you'll see that there
| | 02:30 |
are measures for Overall Success, Overall
Directness, and Time Taken.
| | 02:34 |
As with our card sort, if time taken looks
suspiciously short, then it's likely that
| | 02:40 |
people aren't performing the reverse sort
properly.
| | 02:42 |
But now we'd most likely see low numbers
for the other 2 metrics as well.
| | 02:46 |
Overall success and overall directness are
both measures of how well the reverse sort
| | 02:52 |
is performing.
If these numbers are lower than 50% after
| | 02:56 |
even just half of your participants have
responded, there's a problem with your
| | 02:59 |
information architecture.
You can analyze the results to see whether
| | 03:03 |
the issue is clear.
If it is, then you can close the reverse
| | 03:07 |
sort, make changes to your stretcher, and
then run a new reverse sort.
| | 03:11 |
That means you don't waste participants.
If the reason isn't clear from the
| | 03:16 |
results, you may need to run the whole set
of participants so that you can see trends emerge.
| | 03:22 |
We'll talk about the analysis you can do
on the results in the next section.
| | 03:25 |
| | Collapse this transcript |
| Reviewing what the participants see| 00:00 |
At this point, I'm going to show you what
a reverse sort looks like from the
| | 00:03 |
participant's perspective.
After they click on the e-mail you send
| | 00:07 |
out, this is what participants see in
their browser.
| | 00:10 |
Once they've entered their e-mail address,
they'd be shown any presort questions that
| | 00:15 |
we specified.
We didn't have any presort questions, so
| | 00:19 |
our participants are taken straight to the
sorting interface.
| | 00:21 |
Here, participants see the instructions.
Which tell them how the reverse sort tasks
| | 00:26 |
locations will be displayed.
So.
| | 00:29 |
Let's do the first task.
As you can see, they're presented in a
| | 00:34 |
random order.
That helps make sure that the results
| | 00:37 |
aren't biased by participants getting used
to the information architecture by the
| | 00:42 |
time they complete the later tasks.
Let's start the task.
| | 00:46 |
And I'm finding native landscaping plants.
So.
| | 00:49 |
I'm guessing that would be under live
plants.
| | 00:55 |
And I'd say, I'll find it here.
Then I'm presented with the next task.
| | 01:00 |
If we added any post-session questions,
the participant would see them now.
| | 01:06 |
However, we didn't have any extra
questions.
| | 01:09 |
So that's seeing the final thank you
screen.
| | 01:12 |
At this point, the sort is finished and
the data is captured in the software ready
| | 01:16 |
for us to analyze.
| | 01:17 |
| | Collapse this transcript |
| Analyzing a Treejack reverse sort| 00:00 |
After our participants completed the sort,
we can go back to our Treejack dashboard
| | 00:04 |
and analyze the results.
I'm going to click on the View Results
| | 00:08 |
button for the Hansel and Petal server in
the list of surveys.
| | 00:13 |
We already talked about the Overview and
Participants tabs.
| | 00:16 |
But it's worth pointing out that you can
choose to ignore or even delete
| | 00:19 |
participant data by choosing those options
for a participant on the Participants tab.
| | 00:24 |
You might choose to do this, if there is a
participant who is obviously not giving
| | 00:28 |
the reverse sort of good attempt.
For instance, by skipping most of the questions.
| | 00:33 |
This page also lets you select a sub group
of participants to analyze.
| | 00:37 |
Either manually, or based on answers to
questions you asked them.
| | 00:42 |
That's very useful if you want to find out
whether different sub groups of
| | 00:45 |
participants think about the information
architecture in different ways.
| | 00:48 |
For now though, we'll include all of our
participants in the analysis.
| | 00:52 |
Treejack, gives you the same type of
information that you'd get from a paper
| | 00:56 |
sort, in the form of a destination table.
The correct answers are highlighted in
| | 01:02 |
green, and they've also used a graduated
red tint, to show cells that contain a
| | 01:06 |
certain proportion of wrong answers.
You can use this information to see how
| | 01:10 |
many participants chose the correct
location and which other locations within
| | 01:14 |
the hierarchy they tended to choose.
Treejack also offers some other
| | 01:19 |
interesting statistics and visualizations
that you might find useful.
| | 01:22 |
On the Task Results tab, you can see
detailed statistics for each task, that is
| | 01:28 |
each card equivalent that participants
placed in the hierarchy.
| | 01:32 |
On the left is the overall rating.
This is a combined number made up of
| | 01:36 |
several factors, success, directness and
time taken.
| | 01:40 |
This is an easy way to spot which tasks
didn't perform so well.
| | 01:44 |
The closer to red each entry is, the worse
it was.
| | 01:47 |
Within each task, there's a breakdown of
the individual statistics.
| | 01:51 |
Success is the closest statistic to our
paper reverse sort.
| | 01:55 |
This indicates how many people found the
correct location within the information
| | 01:59 |
architecture hierarchy.
Directness shows the proportion of people
| | 02:03 |
who went directly to this answer, rather
than choosing it after going somewhere
| | 02:07 |
else first.
If you like, it's an indication of
| | 02:10 |
participants confidence in this answer.
And our paper sort this will probably be
| | 02:15 |
the task that ended up on the confusion
group.
| | 02:18 |
Time Taken might be of interest, not in
terms of actual time taken because this
| | 02:22 |
will probably be very different on your
real site, but comparative to other tasks.
| | 02:26 |
If one task took a proportionately long
time, it's just that people won't show
| | 02:31 |
where to to to complete it.
Treejack also gives you one more
| | 02:35 |
interesting visualization.
Pie trees, unfortunately, aren't what they
| | 02:39 |
sound like, trees full of pies.
But they do show you proportion of people
| | 02:43 |
who chose each route for the hierarchy,
and what that action was at each point.
| | 02:50 |
You may find that this helps you with your
interpretation.
| | 02:52 |
For instance, if one of the pies is thick,
It means the participants chose that route
| | 02:57 |
more often.
If one of the gray paths is thick, it
| | 03:00 |
means that many participants chose that
wrong route through the structure.
| | 03:04 |
I would suggest that you choose to view
the horizontal rather than the radial tree.
| | 03:08 |
As I've said before, I really like the
physical nature of running card sorts and
| | 03:12 |
reverse sorts using paper index cards.
However, as you've seen there are some
| | 03:16 |
interesting visualizations that you get
with computer based tools.
| | 03:19 |
Those visualizations aren't essential, but
they may want to help you interpret what
| | 03:24 |
you see in your reverse sort.
The added benefit of opening up your
| | 03:27 |
participant base to a remote audience
often means that people choose to use
| | 03:32 |
these online tools.
| | 03:33 |
| | Collapse this transcript |
|
|
9. From Information Architecture to Navigation StructureGetting to navigation| 00:00 |
Like we said at the beginning of the
course, the term information architecture
| | 00:04 |
refers to how you show visitors to your
site or users of your application the
| | 00:09 |
content you have and the actions they can
perform.
| | 00:12 |
That encompasses much more than just the
menu system.
| | 00:15 |
You can use it to determine how best to
show your content, your site structure,
| | 00:19 |
and even the terminology that you use to
describe things.
| | 00:23 |
In other words, it's one of the basic
tools you'll use to decide how to arrange
| | 00:27 |
your content or data so that people can
find and use it quickly.
| | 00:32 |
So far, we've run a card sort, which tells
us how users grouped tasks on our site and
| | 00:37 |
the category names they use.
We've created an abstract information architecture.
| | 00:43 |
And we've run a reverse sort to test how
well the abstract information architecture
| | 00:47 |
meets with user's expectations.
What these tools don't tell you though is
| | 00:52 |
how to display those information
categories on your site or in your application.
| | 00:56 |
Within reason, you'll want your navigation
menus to be as concise as possible, and as
| | 01:02 |
informative as possible.
In order to get to that point, it's best
| | 01:06 |
to start by thinking about what other
navigation tools are available to you.
| | 01:10 |
Which bits of your structure have a
standard location on most sites, and so
| | 01:15 |
what parts of your information
architecture don't need to be in
| | 01:18 |
navigation menus at all.
After you've removed those, there are
| | 01:22 |
other sections of the architecture that
can exist primarily within the content on
| | 01:26 |
the site, which leaves you with a reduced
number of items that will actually need
| | 01:31 |
navigation menu entries.
In the rest of this chapter, we'll go
| | 01:35 |
through the steps needed to turn your
abstract structure into something
| | 01:38 |
concrete, and then how to check your
concrete designs are working well and
| | 01:42 |
communicate them with other people.
| | 01:44 |
| | Collapse this transcript |
| Standard page elements| 00:00 |
All the way through this course, I've been
saying that your information architecture
| | 00:04 |
is more than just your navigation menus.
Here's where we really demonstrate that point.
| | 00:10 |
In the Hansel and Petal example I've been
using, a pretty strong grouping of items
| | 00:15 |
emerged around support tasks.
Now, help and support is often displayed
| | 00:20 |
as the right most menu item.
but more recently it's been showing up as
| | 00:24 |
a mini menu on the footer area on the
larger and more involved sites.
| | 00:29 |
So there's one part of our architecture
that might not be a menu item at all.
| | 00:32 |
Another is contact details.
Corporate sites often provide a mini menu
| | 00:37 |
in the footer with separate links for
journalists, investors, customers and job applicants.
| | 00:42 |
Again, this produces the number of top
level menu items that are required.
| | 00:48 |
Obviously, for your site design, you might
have a small enough number of categories,
| | 00:51 |
that you decide to keep this as a
navigation menu item, but the option to
| | 00:55 |
remove it is there.
The same applies to about us content and
| | 01:00 |
other administrative information.
Often, this is better off in its own
| | 01:03 |
section like the footer, than within the
main menu structure.
| | 01:07 |
Look at current trends on other sites,
especially ones in your industry to know
| | 01:12 |
whether this is even likely to be
applicable for your site, or whether your
| | 01:15 |
customers are most likely to rely on
traditional menus to find those items.
| | 01:20 |
News style links are another area that may
or may not be wise to put in your
| | 01:24 |
navigation menu structure.
If you're a news related site, your whole
| | 01:28 |
structure will be dictated by news
categories.
| | 01:31 |
But for regular sites where you put news
items will depend very much on the
| | 01:36 |
frequency of updates and the relevance of
the content.
| | 01:38 |
For instance, if your site only releases
news items every couple of months, and the
| | 01:44 |
articles would only be of interest to a
small audience like investors, news might
| | 01:48 |
be something you put on your corporate
page, or your about us page.
| | 01:52 |
If, on the other hand, you have several
items each week that are of interest to
| | 01:56 |
all your customers.
You might reserve a spot on the home page
| | 02:00 |
specifically for this content, with a link
to a news archive page.
| | 02:04 |
Alternatively, news about product recalls
might be best off living in the support
| | 02:10 |
area of the site.
In other words, even if your card sort
| | 02:14 |
participants created a category called
news, it's up to you to decide how to
| | 02:19 |
handle that content.
Does it get it's own menu item?
| | 02:23 |
Does it live within the content on a
specific page, or is it distributed around
| | 02:28 |
the site in relevant locations?
I used news as an example here but the
| | 02:32 |
same questions might well apply to other
categories from your information architecture.
| | 02:37 |
Although they don't replace menu items,
other tools on your site such as
| | 02:41 |
breadcrumbs, In page link lists, and the
organization of content on the home page
| | 02:47 |
and category pages, all contribute to your
visitors' understanding of your content structure.
| | 02:52 |
We'll cover this content based navigation
in the next section.
| | 02:56 |
| | Collapse this transcript |
| Content-based navigation| 00:00 |
We used the card sort and reverse sort to
determine how content on the site should
| | 00:05 |
be categorized.
That categorization will now help us work
| | 00:09 |
out how to display content on pages at
each level of the site.
| | 00:13 |
What order to list items in within the
menu, and how to provide indexing and pagination.
| | 00:17 |
What attributes people are likely to want
to filter by.
| | 00:21 |
Think of your information architecture
categories, as a map to use when you
| | 00:25 |
create pages at each level of your site.
Most sites will have Detail pages,
| | 00:30 |
Category pages and the Home page, along
with some special purpose pages, we'll
| | 00:35 |
discuss in a minute.
I talk about how content should flow
| | 00:38 |
between these page types in my course on
UX fundamental.
| | 00:42 |
It's worth your while checking out that
course for a deeper understanding, but
| | 00:45 |
let's show you now how your information
architecture categories relate to your
| | 00:49 |
content based navigation.
First, what do I mean by content based navigation?
| | 00:54 |
Well, there's navigation at every level of
your site.
| | 00:56 |
From Detail pages up to the Home page.
On your Detail page, you have content
| | 01:02 |
areas, inline links, and related links.
Category pages group several detail page
| | 01:07 |
together with content summaries.
And, you've guessed it, more related links.
| | 01:12 |
Above Category pages, you have your Home
page.
| | 01:14 |
That might contain sections devoted to
each category.
| | 01:17 |
Or maybe only the most important
categories.
| | 01:20 |
It might also contain content, like news
items, that don't really appear anywhere
| | 01:24 |
else on the site.
How do you know what Category pages to create?
| | 01:29 |
With Detail pages to summarize on each
Category page?
| | 01:32 |
Well, that's one use of your information
architecture.
| | 01:35 |
You'll also need to consider the layout of
pages that contain Content Filtering tools
| | 01:39 |
and navigation specific pages, like your
search results page and your site map.
| | 01:45 |
All of these areas benefit from a good
overall information architecture with set,
| | 01:49 |
well defined, well named categories.
For instance, your information
| | 01:54 |
architecture will help you work out,
whether you should split the content on
| | 01:57 |
your Search Results page into categories.
For instance, between people, projects,
| | 02:02 |
and procedures on an internet site.
You can arrange your content based on the
| | 02:06 |
information architecture.
And quickly determine where in the
| | 02:09 |
hierarchy each detail page should live.
Links between pages and between levels in
| | 02:14 |
the content hierarchy should become
apparent just by comparing your content to
| | 02:18 |
your information architecture.
Think of the information architecture as
| | 02:22 |
the guidelines for what content gets
placed on what pages.
| | 02:26 |
I find it helps to draw wide frame
sketches of the different page types, so
| | 02:30 |
that I know what types of content I'll
need to provide on each one.
| | 02:33 |
So for instance, my details pages, I'll
have content, but then I'll also want to
| | 02:38 |
link to related pages, maybe in a right
hand column.
| | 02:40 |
The content may have other components,
such as downloadable versions or videos.
| | 02:45 |
So, where and how will those show up.
At the category level, my wire frame may
| | 02:51 |
show how content snippets from different
detail pages are merged to create an
| | 02:56 |
informative page that contributes to my
navigation efforts, by showing visitors
| | 03:00 |
what content is available to them In this
category, along with a list of links to
| | 03:05 |
other similar content.
With these wire frames, the intention is
| | 03:09 |
not to design the visual presentation of
the site, but rather to show which
| | 03:13 |
elements will need to be on the page, so
that when the time comes to consider
| | 03:17 |
aesthetics,, nothing is forgotten and the
site provides the most information it can,
| | 03:21 |
to help visitors orient themselves and
find the exact piece of content they need.
| | 03:26 |
| | Collapse this transcript |
| Going from information architecture to site layout| 00:00 |
It's only after you've worked out what
parts of the information architecture can
| | 00:04 |
be displayed around all the other areas of
the site that you want to start thinking
| | 00:08 |
about your menu structure.
At this point, you know what things can be
| | 00:12 |
dealt with without needing menus.
So you've automatically limited the number
| | 00:16 |
of different menu items you'll need.
The process for coming up with the right
| | 00:20 |
set of menus is very much like doing your
own card sort.
| | 00:23 |
Do it using index cards, sticky notes, or
sketches.
| | 00:27 |
But do it with your hands, not software.
It's much faster to rearrange things by
| | 00:32 |
hand, so you'll be more tempted to do it
and end up with the best structure, rather
| | 00:36 |
than a good enough one.
Once you have your navigation menu
| | 00:40 |
structure arranged as a set of cards, it
helps to turn it into a diagram that you
| | 00:44 |
can share with other people.
At this point, it's worth putting it in
| | 00:48 |
digital format.
There are several tools that work well for
| | 00:51 |
this, from specialist applications such as
Visio to most drawing applications, or
| | 00:55 |
even presentation tools like PowerPoint.
For smaller sites, you can include every
| | 01:00 |
page in the diagram.
For larger sites, or ones where content is
| | 01:04 |
pulled from a content management system or
database, you might only have to show a
| | 01:08 |
couple of examples for each type in the
diagram.
| | 01:11 |
I've talked a lot about websites during
this course, so I will point out here that
| | 01:16 |
the abstract information architecture
applies as much to applications as it does
| | 01:20 |
to sites.
If the tasks you used in your card sort
| | 01:24 |
are ones that people can perform in your
application, then the groupings they
| | 01:28 |
create indicate how they'll think about
the problem space, and can be used to
| | 01:32 |
develop the controls, dialog boxes, menus,
and terminology used within your product.
| | 01:38 |
That applies as much to a full-on
task-based application, like a
| | 01:41 |
spreadsheet, as it does to a hybrid
content presentation application, like a
| | 01:45 |
city guide.
With your wire frame sketches and your
| | 01:48 |
navigation menu diagram, you're ready to
start migrating content into your new structure.
| | 01:53 |
I don't cover content migration in this
course, but you can watch Patrick Nichols
| | 01:58 |
course on content strategy for more
details on how to perform that task.
| | 02:02 |
How do you know if your categorization is
correct?
| | 02:05 |
Well, first, it should have worked well in
the reverse sort.
| | 02:08 |
80% agreement or more shows that users
understand the information architecture
| | 02:14 |
you created.
The reverse sort should also have
| | 02:18 |
confirmed that your terminology is
familiar to real customers.
| | 02:22 |
You should have fixed any problems and
retested before going any further.
| | 02:26 |
Now, we have the biggest test of all.
It should be easy to assign content to
| | 02:30 |
your information architecture.
You should know where in the structure
| | 02:33 |
each existing piece of content belongs.
And there should be very little need for
| | 02:38 |
content duplication or see also links.
If your information architecture supports
| | 02:43 |
the process of content migration, then you
probably have a sufficiently detailed and
| | 02:47 |
accurate set of categories.
| | 02:48 |
| | Collapse this transcript |
|
|
10. Testing That You've Got It RightThere's no substitute for usability testing| 00:00 |
Once you've applied the new information
architecture to all aspects of your site
| | 00:03 |
or application be sure to usability test
it.
| | 00:07 |
That will let you know whether you've
created conceptual problems, and whether
| | 00:11 |
can find all the places where you expose
the information architecture in the user interface.
| | 00:16 |
Because part of the process involves your
own interpretation, it's important to
| | 00:21 |
check how right you were by user testing
or making tweaks as necessary.
| | 00:26 |
Remember, although you had user research
as the basis for your information
| | 00:30 |
architecture, and although you did a
reverse sort to verify it, you are still
| | 00:35 |
responsible for creating the actual
implementation of that information
| | 00:38 |
architecture on the site or in the
product.
| | 00:41 |
You need to test the design assumptions
that you made during that process.
| | 00:45 |
The easiest way to make sure you've got
things right is to use a traditional
| | 00:48 |
usabilty test, using some tasks very much
like the ones on your inital card sort cards.
| | 00:54 |
If usability test participants can find
the content or complete the correct action
| | 00:59 |
Then its an indication that you
interpreted the information architecture correctly.
| | 01:04 |
For more information on how to ru your own
usability test quickly and easily see my
| | 01:08 |
lynda.com course on usability testing.
| | 01:12 |
| | Collapse this transcript |
| Watch your server logs after you go live| 00:00 |
You should usability test before you go
live, but it's still worth checking in on
| | 00:05 |
your product after you've gone live.
Watch your server logs or instrumentation
| | 00:09 |
to see whether there are specific areas
that people don't go, or a high proportion
| | 00:13 |
of searches for a particular item.
You might need to tweak the information
| | 00:18 |
architecture a little bit in the future.
Your sever logs, search logs, and help
| | 00:22 |
desk calls, will let you know where the
issues lie.
| | 00:26 |
If you start getting an unusual number of
searches for a specific category of items,
| | 00:30 |
or high number of support calls asking how
or where to complete a certain action, it
| | 00:35 |
suggests that either your design needs to
change or that potentially your
| | 00:39 |
information architecture isn't
categorizing items how users expect.
| | 00:44 |
In my experience If you did your card sort
and reverse sort research well, it's
| | 00:49 |
typically an issue with the design
interpretation of the information
| | 00:52 |
architecture, rather than the information
architecture itself.
| | 00:56 |
By analyzing this information, you can
plan and design a change.
| | 01:01 |
And then perform a quick usability test to
see whether your design really helps.
| | 01:05 |
If it does, then you can make a change in
the live product and then track it over time.
| | 01:10 |
It might take several iterations to iron
out all the wrinkles in your new
| | 01:15 |
information architecture.
And in your subsequent designs.
| | 01:18 |
Be patient, you'll most likely see large
increases in users efficiency to start
| | 01:23 |
with, and only minor increases after each
subsequent iteration.
| | 01:28 |
It's up to you how many iterations you
perform, in order to reach a suitable
| | 01:31 |
level of findability for your users.
| | 01:34 |
| | Collapse this transcript |
|
|
ConclusionThe right information architecture is crucial to your site| 00:00 |
Users are happiest when your site's
structure, its information architecture,
| | 00:05 |
matches the way they think about the
problem space.
| | 00:08 |
Without a good site structure, customers
will get lost and frustrated.
| | 00:11 |
Luckily, although customers can't tell us
exactly what they need, tools like card
| | 00:17 |
sorting and reverse sorting, make it
simple to find out how they group
| | 00:21 |
information in their minds.
Then, we can take that information and
| | 00:25 |
turn it into a structure that we can use
on our site or in our application.
| | 00:29 |
I love card sorting as a research tool.
It's cheap, easy to perform, and gives you
| | 00:34 |
a wealth of data about how your users
think.
| | 00:37 |
You can perform and analyze a card sort
with minimal tools and technology,
| | 00:42 |
although there are several applications
and sites out there to help you if you
| | 00:45 |
want to go high tech.
Card sorting and reverse sorting are great
| | 00:49 |
for helping you identify your information
architecture.
| | 00:51 |
Getting early input and feedback from
representative users means that your
| | 00:55 |
information architecture will end up much
more realistic, than if you just tried to
| | 00:59 |
dream it up yourself.
By now, you'll realize that the term
| | 01:03 |
information architecture isn't synonymous
with navigation menus.
| | 01:06 |
Instead, the information architecture is a
map of how your customers view your
| | 01:11 |
content and functionality.
That means that the information
| | 01:14 |
architecture you create will be useful to
you in many areas of site design from what
| | 01:19 |
terminology to use, what to put on
different detail and category pages, what
| | 01:24 |
content to showcase on your home page, and
even how to layout your footers, search
| | 01:29 |
results page, and site map.
The creation process doesn't follow some algorithm.
| | 01:35 |
We can't just throw user data in one end
and pull a perfect structure from the
| | 01:38 |
other end.
There is an art to finessing the
| | 01:41 |
information from card sort data.
The results tell you where to look and
| | 01:45 |
what general groupings people use.
It's up to you to turn that into sensible
| | 01:50 |
clusters of information on your site, and
your participants' comments will help with that.
| | 01:56 |
The results also don't tell you which
items to put in the menu and which to
| | 01:59 |
place elsewhere on the site.
For instance, in the footer, as a content
| | 02:03 |
structuring technique, or in a separate
support area.
| | 02:07 |
There's still a skill involved in creating
a good architecture and displaying it in a
| | 02:10 |
usable and informative way across the site
or application.
| | 02:14 |
There's still a lot of creative work
required and often that will mean going
| | 02:17 |
through several iterations before you get
to a level of agreement and comprehension
| | 02:21 |
that you're happy with.
What the techniques in this course can do
| | 02:25 |
is give you a head start.
It's worth seeing the process through,
| | 02:29 |
because without the correct information
architecture almost every element of your
| | 02:33 |
site will be sub-optimal, causing
unnecessary confusion for your users and
| | 02:37 |
leading to abandonment, rather than task
completion.
| | 02:40 |
So, stick with the process.
There may be times when it's not clear to
| | 02:45 |
you how to proceed.
If that's the case, just go back to your
| | 02:49 |
user data again.
The more you look at the data, and the
| | 02:52 |
more different ways you slice it, the more
likely it is you'll find the
| | 02:56 |
categorization structure that matches the
way your users think.
| | 03:00 |
When you get it, and implement a site or
application based on a great information
| | 03:04 |
architecture, you'll know it not by the
number of comments you get, but rather, by
| | 03:09 |
the lack of complaints.
A good information architecture
| | 03:12 |
implemented well in a site or product just
works.
| | 03:16 |
| | Collapse this transcript |
| Next steps| 00:00 |
Thanks for watching this course.
Designing or redesigning an information
| | 00:04 |
architecture can be a daunting task but
there are tools and techniques that can
| | 00:08 |
help you.
There are also several online resources
| | 00:11 |
and books that you can refer to.
To start with there are several related
| | 00:15 |
lynda.com courses.
My usability testing course will help you
| | 00:19 |
understand how to interact with
participants in a way that doesn't bias
| | 00:23 |
their responses.
And, you should be usability testing the
| | 00:26 |
product that uses your new information
architecture anyway.
| | 00:29 |
My UX fundamentals course describes the
relationship between your different page types.
| | 00:34 |
Detail, category and home pages in a way
that meshes with this course.
| | 00:39 |
There are also some courses that focus
entirely on content strategy.
| | 00:43 |
I already mentioned Patrick Nichols'
course, but you should also check out
| | 00:47 |
creating an effective content strategy for
web sites by Janine Warner.
| | 00:52 |
There are several websites and online
communities that host good information
| | 00:56 |
architecture related content.
Boxes and Arrows and UX Matters are two
| | 01:02 |
sites that regularly publish relevant
content.
| | 01:06 |
(SOUND) The Information Architecture
Institute has resources and discussion
| | 01:11 |
lists that might be useful to you, and for
a detailed description of card sorting,
| | 01:17 |
check out William Hudson's entry in the
Interaction Design Encyclopedia.
| | 01:22 |
William is a big proponent of card
sorting, and created the SynCaps card sort
| | 01:26 |
analysis software.
Finally, the three books I'd recommend are
| | 01:30 |
Information Architecture for the World
Wide Web by Lou Rosenfeld and Peter Morville.
| | 01:36 |
Which is so well known in this field, that
it's just become known as the polar bear book.
| | 01:40 |
For reasons that are obvious when you see
the cover.
| | 01:42 |
And, two books by Donna Spencer.
She writes in an approachable, chatty
| | 01:47 |
style that has great tips for creating the
right information architecture.
| | 01:52 |
The book titles are A Practical Guide to
Information Architecture and Card Sorting.
| | 01:58 |
The first one of these includes a
description of how to use some automated
| | 02:02 |
spreadsheets that can help you with your
analysis and which are a data export
| | 02:06 |
option from OptimalSort.
So, go out and continue your education
| | 02:11 |
with these resources, and then go out and
improve some of the terrible site
| | 02:15 |
structures that exist out there.
Your users will thank you for it.
| | 02:19 |
| | Collapse this transcript |
|
|
