Join Corey Koberg for an in-depth discussion in this video Analysis: Upper funnel and pathing, part of Introduction to Attribution and Mix Modeling.
- [Narrator] As we ponder which methods and models to apply before we jump back into Google Analytics, I want to highlight a few other tools that can help with your analysis of how channels fit into that customer journey. This is the attribution menu from a paid attribution vendor that I like called Visual IQ. You can see that they have a lot of different reports. Some of these are around the customer journey, they have one here around time lag, which we'll look at closer in just a second, and so on. Now, most attribution tools have reports to help us understand that customer journey. But, we also have some of these options available in our free Google Analytics, so let's head over there and continue using that as our example.
Now, per usual, I don't spend much time on these overview reports, but one aspect that I like and some folks find very intuitive is this Venn diagram that gives you a sense for how much overlap there is among channels. Now, I take some issue with some things in the viz, but directionally, I think it can be useful at a high level. In this case we have three different channels. Here we have referral, we have organic search, and we have direct. We can see them on their own, and then we can also see the overlap. So, which of these conversion paths had both a direct and a referral? We see almost 20% here, and then we can also look at the intersection of all three.
So, 2.3% of the conversion paths included direct, organic search, and a referral. And again, we can come over here to the checked boxes, we can add and subtract these to see which one's it in, get these kind of different Venn diagrams, and it gives us a sense for how these things interact, and in a broad sense, what our visitors are doing. Okay, let's explore some other analysis and revisit this idea that some channels are openers, and some channels are closers, and some fulfill that nurture role in between, where folks are researching and narrowing down that consideration set.
Earlier, we looked at what happens when you compare those first click exposers with the last click closers, and the two different models, and we talked about this idea that assists are important, as they set up that close. And this analysis is made even simpler, and perhaps even a little bit better, by the assisted conversions report down here. As I scroll down here, we have our same channels here, and we have the assisted conversions and then the conversion value associated right with that. We also have the next one over here when it's a last click, or these direct conversions, and then the value associated with that.
So, that same comparison we did, which was the openers and the closers, but I've got it right here in the same table. So, the definition of assist in this case is not just the first touch, though. It is any touch that is prior to the last interaction. So, if you have five different touch points, then four of those would be assists, and the fifth one is going to be that last touch. But, in my opinion, the most interesting way to look at this is not the first four columns here, it's actually this ratio that we see in the last column. And we have the times that the channel was an assist in the numerator, and then when it was the last click in the denominator.
So, if the channel tends to be upper funnel, we see a larger numerator, and that's a larger number. If the channel tends to be a closer, then we see that smaller, fractional number here in the ratio. So, since I've sorted by this, we can clearly see which channels function as assists earlier in the funnel, earlier in the journey, and which ones tend to be closers. Now, it doesn't surprise me that display is an upper funnel channel. Most marketers also think social in that regard. They think of this as a way where social can get your brand out to someone who hasn't otherwise heard of it, you can do lookalike modeling, you can use your friends to do those introduce the idea and the brand.
But in this case, we've got social all the way down here at the bottom. It is acting like a closer. It is the one that tends to be in that last click, or right before the conversion position. That's an interesting find. We can dig into the analysis and figure out more, but at first glance, that's a pretty interesting result. Now again, Google Analytics is not the only tool that has this one, Visual IQ here, we have the same kind of thing. In this case, what they're calling true conversions, which is the amount of credit that's being applied to that one, and so they can see what comes here in the conversions by the default, versus what the model is showing for the conversions, and then we can sort by that here, which is the difference, and we can see in this case what that would look like, which ones are going to be openers, which ones are closers, so again we have display here being at the very top of the model, where there's not very many conversions in a last click model, but a lot more being in this kind of assist mode, and then at the very bottom of the funnel we see things like direct, which is probably where we expect to see it.
So, similar kind of report here in a different tool, but works generally the same way. Okay, heading back to Google Analytics, now, we've talked about understanding the journey, but there's a report here that's going to shed a bit more light on this, that's going to be directly what's happening, and that is this top conversion paths. It's going to tell us which ones were involved and how, and even in sequence. So, we can understand how many times they were involved, and what the order was in the top paths that created this. So, in this case, our top one is a referral, followed by a direct. And then, next most common path was an organic search, followed by a direct visit.
Now, I admit this data set isn't particularly interesting, because Google frankly is not using very much media to drive to this merchandise store that we're looking at here. But, I do have a screenshot of one that's a bit more interesting. So, here you can see we have lots of channels. In this case there's lots of campaigns, and sources, and mediums and things that are driving here, so there's ways that we can do analysis on that that can get really interesting. They have a nice, clean data set, and they have lots of sources, so there's plenty of ways that we can look at this and understand in which cases and in which order, which sequence, which ones are happening earlier in the funnel, which ones are later in the funnel, does one tend to follow the other? A lot of people like to look and break down paid search into things like branded versus not, or people originally starting out with those very generic searches, and then moving to the branded search as they start to understand a bit more.
So, we have paid search times two, sometimes if you actually will break that out you can see that as a generic search followed by a branded search. One thing that's worth pointing out here is that unlike the rest of the Google Analytics reports, as we mentioned before, direct visits are going to be considered distinct entity here. You're actually going to see when a direct visit was there. It's not just going to revert to the previous channel. In the rest of the reports, if you clicked on an ad and then came back direct and converted the ad would get the credit. But here, since we're looking at this in very fine detail you are going to actually see when those direct visits are going to play a role.
In the previous video we showed if we want to remove those how we can do that, but by default they're here. It's also worth pointing out that direct visits sometimes can be a goal in and of themselves. If you have a new brand or product, just getting people to know your brand and your site well enough to specifically choose to visit your site just by typing your domain name in from memory, that may be a win in and of itself. That could be part of the goal. Now, if you're curious about the number of touch points that customers are taking, you can come down here to the path length report. This is going to give us some insight on how many touch points we have.
Do most folks tend to buy on their first visit, or is this something that requires research and contemplation? This report, which can and should be segmented, can really reveal some interesting trends about how your media and your content is consumed. You also have the time lag report. Remember, that was one of the ones we saw on the other tool as well. This one can be interesting from a journey analysis perspective to understand how long in calendar days it takes folks to convert from the time they first interact with your brand. Now, this report helps us do that. Some caveats apply.
These are going to be sessions from a single device, so it's only going to capture the media and the clicks from that device, so we're generally talking about digital only here, and a single device, but it can still help us understand given that constraint how long in time are people taking to go from first interaction to conversion. I encourage everyone here to learn as much as they can about your customers' online journeys, and these reports can definitely help with that analysis.
In this course, marketing expert Corey Koberg dives into the basics of attribution and mix modeling. Corey explains what the models are and goes over a few of the most common ones. He shows how to approach offline data, goes into how attribution modeling and marketing mix modeling (MMM) work together, and shares best practices for using different attribution models. Plus, he walks through the challenges you may face when you start analyzing marketing data with models, and goes over an action plan that can help to ensure that your deployment goes as smoothly as possible.
- Determining why attribution is important
- Reviewing last click, first click, linear, rule-based, custom, and data-driven models
- Deciding when to use MMM or multi-touch attribution
- Considering challenges
- Evaluating and deploying