Lander teardown: 14 variants, the one that won was the ugliest

Ran 14 lander variants on the same sweepstakes offer, same traffic source, ~$9k total spend to get clean data.

The winner was a plain single-column page with a system font, no hero image, and a visible form on load. It beat the designed version by 31% on CVR.

My read: the designed page looked like an ad and the plain page looked like a form you were already halfway through. Perceived effort is the conversion variable people never test.

Before anyone asks: yes the offer and traffic were compliant, and no, I am not posting the offer.

Perceived effort. That is the right phrase for it and I am stealing it for client calls.

Confirmed on push traffic too. Every time I have let a designer near a lander, CVR drops and the page wins awards internally.

Does this hold for higher-consideration purchases though? Plain pages feel risky for anything over $100.

Good question and no, it does not. Above about $80 AOV trust signals start paying for themselves. This is a lead-gen finding, not a universal one.

What tool are you using to split at the lander level without tanking page speed?

$9k to get clean data on 14 variants is the part beginners skip. Most "split tests" posted here are 200 visits and a vibe.