GTM Sauce

Sizzle AI · B2C · AI learning app

Test clean, warm and social-proof sign-up screens; break ties with assisted conversions

Social proof: more purchases in some groups, neutral overall on assisted conversions

Mixedonboardingpaywallexperimentationanalytics-attribution

What they did

Run big variants first, then micro-variants within the winner, on a weekly cycle: three registration pages (cleaner, warmer, more social proof) and paywall variants (free vs paid plan comparison, more vs less detail). Track purchases and engagement together, since Sizzle wants students to actually learn and will pick a hybrid that doesn't cut engagement. Track assisted conversions per screen (screens that don't convert themselves but raise later purchases) as a tiebreaker when purchases are neutral.

What happened

Social proof raised engagement, and purchases in certain demographics or pathways, but not across all users; on assisted conversions it was neutral, so not the harm the top-line number suggested.
Stage
growth
27:49
“rather than the test itself where it's three types of registration, one is more cleaner, one is warmer, one is there's like more social proof, tracking all of these metrics, same thing then on paywall side where it's just trying different variations, whether it's comparison, free versus a paid plan, more detail versus less detail.”
Yev Marusenko
29:44
“If purchases are neutral in your test, look at assisted conversions. That will help deciding some of these screens helping or not.”
Yev Marusenko
“Purchases were up, but in certain types of demographics or pathways, but when you look at it as a assisted conversion, then it was like neutral”
Yev Marusenko

Related topics

Get tactics like this every Monday

New tactics from the week's founder interviews, each linked to where it was said.

More from this episode