I used the landing page and waitlist invite. Ads/intros can be pretty similar as you would do with the actual product. Change call to action from “Sign up” to “Join the waitlist”.

Question for me was (and still there) is where to get as much traffic so the stats become credible enough for making inference.

Didn’t Eric Ries get traffic on adwords for like 5 cents a click or something? If you had funding you could buy cheap targeted traffic and run experiments on that back in the day, but these days traffic is a lot more expensive and a lot more competitive.

Yes you should be running experiments, but no you shouldn’t worry about that too much unless you have enough traffic to actually run meaningful experiments.

You should talk to as many potential customers as you can but.

I think the ideas still apply. I have a b2b software startup in a niche market so obviously I can’t rely on data. Still there is so much you can do. I sold the product with a frontend-only prototype. We had signed letters of intent before we had the prototype. We interviewed over 50 potential clients.

For user experience we do live video recording of clients using the app. We also monitor login behavior etc .

The key is that you need less than you think. I know a guy who signed his first 100k deal based on a promise.