A weak icon at launch makes every install cost more, for months. StoreLab splits real ad traffic across pixel-accurate App Store and Google Play pages and tells you which variant to ship, before launch day. And it lives in your agent: ask Claude for a test, approve the publish, ask who's winning.
Google's Store Listing Experiments and Apple's Product Page Optimization only work on a live app. Your launch traffic, the most expensive traffic you'll ever buy, is spent before they can tell you anything.
Geeklab starts at €300/mo for one test a month, with consultants bundled whether you want them or not. Others don't publish prices at all; you get a sales call.
Icon picked in Slack, screenshots from the last build, validated by real UA budget. StoreLab exists so the guess never ships.
Every StoreLab feature is a tool your agent can call over MCP. Fourteen tools cover the whole run: draft, creative, publish, results, decision.
The whole demo above is real tool calls. Nothing shown is a mockup.
This is how a StoreLab test runs. Each ad click routes a new visitor to a different variant, so the store page changes as traffic splits. Tap around, hit install, and watch every tap roll up into the decision. Or press play and let it run itself.

Test one thing or the whole listing. Pick an element to see the change that moved install-intent, and by how much.
One experiment, a tagged link for every platform. StoreLab bakes the UTM source into each one, keeps the same variant split everywhere, and breaks install-intent down by channel, so you know exactly where your budget converts.
Every experiment resolves to a decision card a founder can act on. Behind it: honest measurement you can defend.
Your agent's access follows your plan. Upgrade mid-session and the new tools work on the next call, no reconnect.