I'm assuming the bad advice is the bit about not reading sites like useit, rather than the "get out there and think" part. But I guess I see it as good advice

I'm assuming the bad advice is the bit about not reading sites like useit, rather than the "get out there and think" part. But I guess I see it as good advice that was communicated poorly, since I agree with your response here. Yes, you absolutely have to watch people use your site. You need to watch them use your competitor's sites. You need to watch them use sites that come up in conversations about your products. You do not need hundreds of people to make a difference. Get people off craigslist and other classifieds for $25 and have everyone that touches the product watch them use it (as video, after the fact). This input is hugely valuable. Metrics are also very important. If a button is on a page in three places, track which one they click. Measure response times, page flows, all that good stuff (although be wary of blindly conflating marketing success with usability). If these are the lessons you've learned from Neilsen, I can only say I'm surprised because I see his work as the opposite approach. Perhaps because I've seen him cited so often over the years by people who don't think they need to do this because they think Neilsen already answered that question. Nobody I know that actually embraces this kind of hands-on testing has any respect for his "studies". What I'm trying to say is don't try and distill universal rules from this narrowly-scoped input. Don't ask a small number of people about a small number of experiences that have no logical connection, and try to infer some wisdom from it that you can apply to another group of people in another experience that may not be at all similar. I've had a case where adding clearer text to a vaguely worded button killed performance. Others where making a button look considerably "worse" improved success dramatically. Sometimes you need to make the font bigger, or add an icon, or remove an icon, or have a popup, or add two pages to your wizard, or not even have a wizard, or whatever. Sometimes you need to do things the wrong way because you lose from inconsistency what you gained from local quality improvements. But usually just need to do things the way that makes sense to you, learn from your mistakes because you fully understand the context, and do it better next time.