How customer development prevented another failure

this is really flawed logic. i hope no one takes this advice. including todd. here is sample size calculator - http://www.surveysystem.com/sscalc.htm

if he had 100 clicks and 3 buys, then his margin of error with 95% certainty is +/- 3.4% - maybe that's enough to make a difference?

but more importantly, companies like twitter never happen under this methodology. i guess this may be sort of ok to use for incremental products, but nothing big will ever be developed this way.

I find CD to be very dangerous. Let's say you have a decent idea. After creating the landing page you toy with 300 dollars worth of adwords, facebook ads and other media buys. Lets say you end up with 3000 visits. The problem is that you are not likely to get the right sample with that small amount of visitors. You probably need to go through a multitude of ad variations and segmentations to find the right market.