Taking a Product Thinking approach to AI
At DataSparQ we bring a product mindset to AI and believe that Product Thinking is vital if you are to introduce AI to your business in such a way that it becomes a sustainable and impactful part of your value chain. Here are our tips for keeping a product mindset when looking at AI for your business. These approaches are designed to help maximise the chance of success for your AI project.
Love the problem, not the solution
AI and machine learning are solutions which are easy to love. They are futuristic, they can surprise and delight with their output, and can produce results which feel like magic. You can quickly become attached to your latest machine learning model or AI solution.
But the solution itself can become a distraction from the real problem at hand. There’s little value in a brilliant AI solution to a low value problem or a problem that could be solved more easily and cheaply through a more traditional approach.
Our 3-phase approach to introducing AI to your organisation (Identify, Qualify and Validate) starts with truly understanding the problems that need to be solved and the value that can be released by solving them, before deciding if and how AI is an appropriate solution.
Stay user-focussed
No AI or machine learning model runs in isolation. Somewhere there is a touchpoint back to the business and a real person. It might be through a dashboard, or a recommendation, or a result which allows another action to be taken. What’s more, some solutions may have more than one interface to the business. For example, the output of a churn prediction model maybe used:
- by operators in a call centre, so they know the likelihood of churn of each customer who calls in
- by a CRM system to generate emails with retention messages for marketing
- by the leadership team to understand global drivers of churn
An AI solution that doesn’t consider the user can end up being ignored, or worse still actively discourage or annoy the user.
AI can only be successful in delivering sustained value if it’s used and becomes a core part of your business processes. It needs to augment the people in your organisation to allow them to perform better. All the successful AI implementations we see and have delivered have this in common — they’re loved by their users and seen as a net benefit, not a threat.
Know your metrics
As with any product, there needs to be a way to identify if the problem you set out to solve is being solved by your solution. The metrics you use for this may be the same that highlighted the problem in the first place, or you may need to implement new analytics in order to track your KPIs.
If your AI solution isn’t having an impact on the KPIs then it’s not working and needs to be retrained, or re-designed. The best AI solutions will have an element of learning so can optimise towards defined goals, but without attention they can fall into the trap of optimising for local maxima. Make sure you’re measuring the right things, and be ready to shake things up (for example introduce a new champion challenger using a different ML method) when improvements look to be reaching a local peak.
We have developed a canvas which helps you bring clarity to the key elements of your AI solution. By collecting your ideas and thoughts in one place you can more effectively validate, iterate and share with others in your business. Get your free copy of the Product Primer Canvas here.
If you want to understand more about how to get started with AI and machine learning in your business, sign up for our free webinar on 25th June: How to kick-start your AI journey.
