At Intercom we’re constantly re-evaluating our process for building great product. Asking our- selves what the best process is for building product people find valuable and useful, a prod- uct they love. One thing we place a huge emphasis on is research. We hire people with direct research experi- ence, and everyone on the product team talks directly to customers: the product managers, the designers, and of course our research team. We also hired a Director of Research (Sian Townsend, who previously led the research teams for Google Maps) much earlier than most other startups. While it’s obvious you should be talking to customers frequently to try and understand their needs, it’s not obvious what the best tool to do that is. We constantly think about things from first principles, and so very early on we applied that lens to how we were talking to customers. We were big fans of the Jobs-to-be-Done framework, but most of what was written on Jobs-to-be- Done was applied to milkshakes and chocolate bars. There was little published research on how to apply Jobs-to-be-Done to software. So we created our own process based on what we knew. Over most of my career I used personas and scenarios to understand customers. Popularised by Alan Cooper in The Inmates are Running the Asylum, they have become one of the most wide- ly used tools in a research and design team’s toolkit. Cooper also wrote a fantastic book called About Face, and I recommend it to all designers who join my team. But I tell them to skip the chapters on personas. When I worked at Google, I created dozens if not hundreds of personas across many projects. We followed Cooper’s method carefully, along with iterations of our own. Universally, I found their value limited. They often helped with building empathy amongst employees who were disconnect- ed from their users, but they rarely led to breakthrough ideas or fresh thinking about a problem. And we have never used personas at Intercom. The first time I really started to question the value of personas was when I left Google and joined Facebook. Facebook has amassed an incredible quantitative data set about what people actually do with their product, and talking with the data science team never failed to educate me. One of the striking things about this data was how similar people’s behavior was. Personas had led me to believe that people are really different, with really different goals. But the similarities were far greater than the differences, and across everything you can imagine – race, age, gender and so on. For example, the motivations behind a married mother of three in the USA posting photos of a family barbecue are basically the same as a Korean teenager posting photos of the house party the night before. The goals and attributes look totally different, but their motivations are the same. Personas would never lead to the same product being designed and built for both these audiences. While best-in-class personas focus on goals (what drive people’s behavior) as well as attributes, the reality is that most personas focus on attributes alone, even when goal driven. 33