Brilliant. The gem for me is the statement that a ton of small decisions creates a defensible moat perhaps even more formidable than adding features. (Biased opinion: After 20,000+ hours of work, Stormpulse is an example of this.)

Yeah.

When I built delicious there were literally hundreds of competitors, many starting as literal clones. Because they didn't understand the reasoning, they couldn't make the same decisions that I could.

I think that every successful product requires a strong vision. Start out small (MVP) and build on that vision. If you follow that path implementing "small" user requested features will most of the time be straightforward and will actually build on your initial vision of the final product.

However, sometimes you have to ignore some requests (big or small) if they aren't in line with what you want the product to eventually be.

One last thought is pivoting - which only select companies have managed to pull off - and have profited incredibly (Flickr, Groupon, PayPal). While it's not necessarily on topic, it's important to listen to user feedback and look for potential in every single idea (no matter how silly they might be).

What if no one ever asks for the most important feature? We get a lot of feature requests, but the number one feature we focused on: "When the map loads, is it easy to know where the nasty weather is?" No one ever asked for that, but it's the killer feature.

"If I’d asked my customers what they wanted, they’d have said a faster horse" - Henry Ford

Indeed. Users rarely ask for what they need, but they do ask for what they think they need. The challenge lies in translating what they asked for into the actual problem they had and then solving THAT.

I think the thing is that you get the feature done, and then ask the users how to polish. You have to come up with the feature yourself.

It's up to you to constantly look for "killer" features, but sometimes user feedback might point you in the right direction.

Chris Anderson (Wired editor) in his book "Free" had one strategy - to never write any feedback down on paper. By never writing it down, you only remember the things that get repeated by your customers frequently, and you just forget everything else.

It's kind of like a high pass filter for feedback, with a very cheap implementation cost ;)

Obviously this tip won't suit everyone all of the time so don't flame me. I thought it was an interesting idea though. It's a really good book, btw; some good advice and stories if you are building freemium web apps.