I think when Eric talks about taking out Google Adwords for testing, he means to do it in the way that "The 4 Hour Work Week" by Timothy Ferriss uses it. He uses Google Adwords as a mechanism to test his ideas and lays out certain criteria to measure success.
37 Signals already had a product when they bought the adwords. These guys are talking about buying a small amount of adwords and using a fake signup page to test a need, before the product even exists.
Right, sorry I wasn't clear, I was only using their more recent adwords success (or lack thereof) now to extrapolate what their success would be on "buying a small amount of adwords and using a fake signup page to test a need, before the product even exists.".
If adwords isn't so good for them now that they have a fully functioning product and well known brand, can you image how much less successful running an adwords campaign would have been for them 3-4 years ago before Basecamp even existed.
It would have probably sucked, and if Jason only listened to this hypothetical adwords datapoint would he have said, "you know what, our adwords account isn't driving anyone to our fake signup page, we better not build Basecamp"?
37 Signals already had a product when they bought the adwords. These guys are talking about buying a small amount of adwords and using a fake signup page to test a need, before the product even exists.