All products have a core experience. This creates the fundamental connection between the user and the product — the hook. For example, on Netflix, this is watching a movie or a TV show. On Amazon, this is buying a product. In a game like Candy Crush, this is playing through a level. All other activities and features in the product support this core product experience. Without the core, the product no longer exists. Each instance of a product might affect users differently. For example, the starting configuration of a Candy Crush level changes the user’s core product experience. If there are many such possibilities, product teams think deeply about what possibilities should be shown to the user. In most products, product design means a product designer/manager type person coming up with a set of hand crafted rules for how a product should behave. ‘If the user has not opened the in-game shop yet, give them a free booster when they reach level 5’. The number of rules explodes exponentially with the number of possibilities for how a product can behave. In such cases, personalizing the core product experience results in a superior user experience, which ultimately increases engagement and retention. Machine learning can help here. Machine learning works best when there is high value in personalization and a large number of possibilities for product experience. ML can make better product decisions, if we know how to use it well and integrate it deeply into our products. Making product experience decisions with ML Let’s think about Facebook. The news feed has a set of machine learning models that ranks the most relevant content to show, drawing from all available content on Facebook. News feed tries to posts that you might comment on, videos that you will watch, ad revenue etc. The behaviour of these News Feed models form your perception of Facebook the product. We’ve all had instances where we see something in our News Feed and say ‘WTF, Facebook, why are you showing me this’. Personally, I have never said ‘WTF, News Feed ranking model, why are you showing me this’. balance multiple criteria , The News Feed machine learning models affect the core product experience of Facebook. The News Feed product teams make decisions that affect the ML models, and the models then decide the product experience for the user. ML is now a weapon at the product manager’s disposal. This is used to indirectly affect the News Feed product experience to achieve product goals and improve metrics like news feed video engagement. When ML is used to drive core product experience, the principles of product management remain the same. However, since product managers work through ML to achieve product goals, new techniques are needed. More articles on Machine Learning and Product Experience: How to build a dynamic garden using machine learning Whoa whoa! ML knows what’s fun now? AI, humanity and the future of product design Machine learning: preventing unintended consequences At , we believe that all products in the future will be dynamic. We’re building a platform that redefines product management and user experience, starting with dynamic difficulty in games using machine learning. Product ML
