Developing and presenting the business case

- [Instructor] The purpose of this meeting is to validate the business reasons to kick off the AI project. Going back to the e-commerce site example, let's say that this is the primary business of the organization, so the project proposal could be including a recommendation engine that takes into account the current shopper's previous purchases, what they're likely to buy based on their profile, and what's currently in their shopping cart and to suggest products that are complements to it. Now the new project proposal comes from the product owners. The product owner understands the business domain, and they have the respect and trust of the business, so they make decisions and determine what the priorities are. The product owner doesn't have to be very technical, but they do need to understand which technologies would benefit the business. They need to know where competitors are using AI, and the industry usage. Now normally, the business purpose of an AI project falls into three broad categories. So there's revenue growth, cost reduction, and reducing risk. Revenue growth is all about selling more products and getting more customers online. Reducing costs might include things like reducing the spend on marketing or advertising. And finally, reducing risk is about things like health and safety and regulatory compliance. So projects that are credible, and have a good chance of increasing revenue growth, are a much easier sell to the management team than projects about reducing risk. Health and safety and regulatory compliance are stuff that companies just have to do. Even if they do a really good job at them, it's unlikely to earn them more money. With many AI projects, the challenge is championing the value of something that might initially be a cost to the business. Now the product owner and the data scientist would've worked very closely together to create the business case. This business case will be presented to the management team. They might've worked through the AI assessment, and determined which product was small enough, but achievable, to introduce to the business in a couple of weeks. So the data scientist would've helped the product owner understand what a minimum viable product would look like, and what it would be able to achieve. They would've helped the product owner estimate how long this project might take, and what resources are required. The data scientists would have the skills to take the current business data, pre-process it, so that it can be fed into the AI model, and develop the AI model. At this stage, it is also helpful to provide a broad overview of what success looks like. If the AI product is substituting or complementing an already-existing process, you can use the current process statistics as a baseline. For example, with an online e-commerce site, it might be possible to show how many additional sales have been made because of a recommendation engine versus the current process. As part of providing justification for this project, they might provide an industry view to the management team. So what are other competitors doing in the market, and include other frameworks, like a SWOT analysis. Depending on the organization and the relevancy, providing a view of the AI assessment might also be helpful to provide a business vision to the leadership team. The product owners would also provide an estimated cost of the project, the resources required, and how long they expect the project to take. Now just as the product owner will propose the AI project to the business owners or the management team, the business owners are responsible for funding the projects. The questions that the management team want the answers to fall typically into two categories, why and what. So why should they fund this project? Is it a good investment? What's the predicted return on investment on this project? There'll be other projects that are competing for funding. How is this project any different, or what's the likelihood that this project will succeed? Or what difference can this project make to the revenue of the business, and what's the cost to the business in terms of finance and resources? The biggest chunk of time when presenting a business case is the product owner providing the proposal, and then understanding and answering the why and what questions from the management team. The other important questions that we need to answer is the how and the who. This normally isn't the focus of this meeting, but it's often helpful to provide a high-level view to the management team. Often what happens is that once a budget has been approved for the AI project, then a separate meeting is organized with other stakeholders. Now because you're trying to make AI a part of the organization, I think a high-level view of how you go about this is possibly beneficial to the management team. So how will the solution work, or what data will be used and how will it be accessed? Or what additional hardware and software resources are required? Now the current platform owners are an integral part of the implementation team. They could be the guys who oversee the team that currently manage and own the current IT platform for the business, and are responsible for it. In the e-commerce business, then the platform team understand the e-commerce site better than anyone else, and manage and maintain it. If changes need to be made to the e-commerce site, then they would know how to make it, and these guys are probably the ones who put the e-commerce site into production in the first place. Now, depending on the organization, getting easy access to data can be difficult. Now if there's management buy-in at this early stage, then the platform owner and the team can ensure that the data is easily accessible. In reality, the product owner and the data scientist would've spoken about this project to the platform owner weeks in advance, and would've normally got access to some of the data to run experiments and see how feasible the project is. As part of making the solution go live, the product owner and the data scientists will require resources from the platform team for the duration of the project. They'll need buy-in from the current platform team in terms of the solution and resourcing, as typically the platform team will be responsible for deploying the solution into production. Now one other question that will need to be covered is when is the project scheduled for? So how long will it take? A scrum master will work closely with the product owner and the development team. They will help make sure that the project is delivered on time and to budget. Now, typically, the scrum master would've been involved in IT projects for the business before, so they know what is involved in such projects and how realistic the timeframes are. The product owner and the data scientists would've included the scrum master in earlier discussions, determining the time scales and the resources for this project. The scrum masters are also there to answer questions from the management team about the project and the time scales. Now remember the purpose of this meeting was to validate the business reasons to kick off the project. From the perspective of the product owner and the data scientist, it is to get funding for the project and the required resources from the platform owner. If the business case isn't accepted, the product owner and the data scientists need to get a full debrief from the management team to understand the reasons for the decision. There could be some very legitimate reasons why an AI business case is rejected, and the product owner and the data scientists need to regroup and review their proposal.