← Adam Garceau

Surveys are mostly BS. Here's what I use instead.

People tell you what you want to hear. Money doesn't.

I don't trust surveys much, and I say that as someone who runs them constantly.

The problem is people lie on them. Not always on purpose. They game them for the reward. They tell you what they think you want to hear. They say they'd pay for something and then never do. Depending on who you're asking, the honesty drops off a cliff. So you end up with a spreadsheet full of confident numbers that don't predict what anyone actually does.

Real answers come from one place: people voting with their wallets. Someone pulling out a credit card is worth more than a hundred people checking "very likely" on a form. That's the only test that fully counts, and nothing replaces it.

The catch is that the wallet test is expensive and slow. You have to build the thing, launch it, and spend real ad dollars to find out. For a long time that was the only honest way to learn.

So here's what I actually use now, in front of that. I build a synthetic audience out of real customer language and run the idea past it first. It's not the truth. It's a filter. It kills the obvious losers, surfaces the objections I hadn't thought of, and ranks my options, in minutes, for almost nothing. Then I take the survivors to real people and let the wallets settle it.

The difference from a normal survey is who's answering and why. A synthetic audience built from real reviews and call transcripts isn't trying to please me and isn't chasing a gift card. It just reflects what the research already said, faster than I could read it myself. And on a real project, the two packages it predicted were exactly the two real customers went on to buy.

Cheap directional read first. Wallets last. Skip the part in the middle where a survey lies to your face.

I'm a marketing director who builds his own tools. More at adamgarceau.com.