User research is, by definition, bespoke. Unlike other workflows where the output is relatively consistent (e.g. a Figma prototype that is handed off to engineering to begin development), research output can refer to unified product feedback, a detailed evaluation of the core user base, or even a recommendation to change product strategies.
For this reason, we heard consistent excitement about how AI can automate the more laborious parts of the user research process, including setting up studies, recruiting users, personalizing surveys, and aggregating insights. It’s early innings here, but companies like Maze and Dovetail are already helping to drive this change.
User research is only as good as the quality of its sample size. Collecting feedback from what is truly a representative sample of a company’s customer base is incredibly difficult to do. AI provides for greater personalization at scale: Instead of asking survey participants to complete the same survey, what if every survey was customized to each user based on their prior responser history, product usage patterns, and current pricing plan?
In the future, user research teams will have access to more diverse forms of high-quality data as AI becomes more mainstream. For instance, the multimodal impact of AI across text, audio, and video may change the way we gather user research significantly in the coming years... For example, many users may not want to fill out a 5 or 10 minute post-purchase product feedback survey, but what if they could leave a 30-second audio review of the product? Already, AI tools can synthesize this type of feedback automatically and incorporate nuances in characteristics like tone and expression.