Launched this week
Versuch.ai
Web Analytics - Find out why they don't pay, then fix it
8 followers
Web Analytics - Find out why they don't pay, then fix it
8 followers
Versuch is the all-in-one conversion stack for early-stage startups: heatmaps, session replay, on-page surveys, A/B and pricing tests, and feature flags from one script tag. Every result is judged by real revenue (Stripe, Shopify, WooCommerce), not vanity clicks. Built-in AI finds the test worth running, writes your surveys, and reads the results back to you. Bayesian stats give a clear chance to win from your first conversions. One line to install, live in minutes.
Really like what you've built here :) A couple of questions. First, your integrations (Stripe, Shopify, WooCommerce) are all setups where the revenue flows through your own account or store. We use a merchant of record instead (Paddle) where the payment and payout data lives on the provider's side. Do you support merchants of record at all, or is there a way to send a conversion value from our own frontend or import revenue data manually? And second: for low-traffic, early-stage sites, how much traffic does the Bayesian model typically need before it will call a pricing test with any confidence?
@alieksia Hi! Thanks for the great questions.
I'm working on adding native paddle integration, however the current analytics api available you can manually update via a webhook or sdk on your own events.
On the bayesian model, we won't call a winner or loser under 100 exposures per variant. Above that, a variant is declared a winner at ≥95% chance-to-beat with a negligible expected downside. We do also have a frequentist engine if you prefer classic significance testing.
@versuch Thank you, that's exactly what I hoped to hear :)
Finally a tool that judges A/B tests by actual Stripe revenue instead of just clicks, that part sold me. The one line install and live in minutes claim actually held up when I tried it on a landing page.
@emeldemirkollu Thanks! Glad to hear it