Thanks, a good write-up! The concept of "demand generation" is completely misleading. Demand cannot be generated. It exists already, and one just need to find it and tap into it. If it doesn't exist there is no product opportunity.
I must comment on one sentence though: if someone's “discovery interviews ask users what they want” then they are incompetent at discovery. 😅 Observing users’ work has always been the user research method that generates the most interesting insights most reliably.
I like the framing of this one. I have definitely found the patterns some of the companies like Anthropic is using to decide on features to go to market with to be interesting. As a user, I’m both shocked on how rapidly they build to fill a problem and also at times feel a little overwhelmed by the amount of change. Like…am I going to spend so much time trying the new things that I never really realize the value from building a system like this.
Thanks Mike - completely feel that tension. I've scrapped about three attempts at building systems with earlier models because the ground kept shifting underneath me. To me, it finally feels stable enough now to grow things incrementally rather than starting over every few months. What do you think would change that for you - the pace slowing down, or just forcing yourself to commit to one stack?
I've found myself doing experiments, adding new things each step along the way to what I've already had working.
I will dabble a little bit here and there when I hear a new product/feature is coming that might connect into my workflow. I'll test it in isolation to see what I think, and if it's cool, I'll give it a run, but I won't just scrap the entire thing. It's like building more... agnostically.
The hardest part had been, for a long time, just remembering all the things I was trying to do in each feature or app I would build. But even using basic things like writing lessons learned or iterating on MDs has helped significantly more than most of the random new things these companies offer.
Building agnostically is a great way to put it. The isolated testing practise is something I'm doing too - it's way too easy to make one wayward decision that messes up everything you already had working. And yes, simple markdown for tracking decisions and refining has been surprisingly effective for me too.
Thanks, a good write-up! The concept of "demand generation" is completely misleading. Demand cannot be generated. It exists already, and one just need to find it and tap into it. If it doesn't exist there is no product opportunity.
I must comment on one sentence though: if someone's “discovery interviews ask users what they want” then they are incompetent at discovery. 😅 Observing users’ work has always been the user research method that generates the most interesting insights most reliably.
I like the framing of this one. I have definitely found the patterns some of the companies like Anthropic is using to decide on features to go to market with to be interesting. As a user, I’m both shocked on how rapidly they build to fill a problem and also at times feel a little overwhelmed by the amount of change. Like…am I going to spend so much time trying the new things that I never really realize the value from building a system like this.
Thanks Mike - completely feel that tension. I've scrapped about three attempts at building systems with earlier models because the ground kept shifting underneath me. To me, it finally feels stable enough now to grow things incrementally rather than starting over every few months. What do you think would change that for you - the pace slowing down, or just forcing yourself to commit to one stack?
I've found myself doing experiments, adding new things each step along the way to what I've already had working.
I will dabble a little bit here and there when I hear a new product/feature is coming that might connect into my workflow. I'll test it in isolation to see what I think, and if it's cool, I'll give it a run, but I won't just scrap the entire thing. It's like building more... agnostically.
The hardest part had been, for a long time, just remembering all the things I was trying to do in each feature or app I would build. But even using basic things like writing lessons learned or iterating on MDs has helped significantly more than most of the random new things these companies offer.
Building agnostically is a great way to put it. The isolated testing practise is something I'm doing too - it's way too easy to make one wayward decision that messes up everything you already had working. And yes, simple markdown for tracking decisions and refining has been surprisingly effective for me too.