Great post, Ross. Let the customers drive the product — after all they're the ones who will use it. (Well, you do need to build the starting point yourself, or there's no baseline for them to think against.) I do think there are some opportunities to anticipate customers' needs, however, and there's one particular case I'd like to mention. Sometimes, instead of adding individual features that customers have indicated a need for, you can add a general "super-feature" that provides a range of individual features all at once, covering what's known to be needed but providing more as well. To pick an extreme example to illustrate, if your product is in English, but then someone wants a French version and someone else wants a Spanish version, don't create French and Spanish versions but instead make the product multilingual, which not only covers the requirements but also covers any future languages. Which there are likely to be: there's a saying in French that "jamais deux sans trois", meaning "never twice without thrice". The principle applies for less-obvious stuff as well, like when I was building a product for banks: one would invent a special kind of floating-rate note while another would invent a special kind of interest-rate swap. Instead of adding each one individually as a feature, I'd look to see whether there was a general approach that would cover both. It would often be a lot more work to implement, but it would automatically cover similar needs that came up in future, with zero wait for the customer.