Funnel Wiring for App Analytics is where we found the double-fire on checkout_start. I came for the flagship design course; I stayed for the taxonomy fight in week two of Wiring.
Reviews
Readouts from people who sat in the room.
We keep glowing blurbs off this page. What follows is how alumni describe specific modules — including the parts that pinched.
Short note: the holdout lesson in Power, Peeking, and Guardrails is the one I still quote in stand-up.
★★★☆☆
Materials are careful. The Variant Atelier pace assumes you can protect Wednesday nights. I missed two critiques during a store emergency and felt behind for the rest of the term. Recordings exist on Hypothesis Desk; live seats do not rewind.
I am not a statistician. Cohort Readout after Mobile Tests finally gave me language for “this slice moved, the base did not.” Our CPO still wants a single percentage; at least I now refuse to invent one.
★★★★★
Nalinee’s randomisation-unit drill in Experiment Design for Mobile Releases is unkind in the best way. We had been assigning by advertising ID on a kids’ learning app. That sentence still makes my stomach drop.
Case notes
A flat result that still changed the release train
A Variant Atelier squad ran a three-variant paywall on Android only, with crash-free sessions as a veto guardrail. Conversion did not move past noise. Unsubscribe on the annual plan ticked up on variant B. They shipped nothing, and they wrote the first “we will not retest this surface until Q2 instrumentation is clean” memo the company had ever circulated. The win was the veto, not a lift.
Limitation they still own: iOS was left out because StoreKit sandbox time slipped. They returned later via Release Consortium hours.
Store screenshots that looked “clearer” and hurt activation
A Hypothesis Desk student used Variant Craft methods on Play Store screenshots. The “cleaner” set won on tap-through from search and lost on day-1 activation. They rolled back within the Play console window and kept the older, busier screenshots. The memo now lives in our archive (anonymised) as a warning against using listing CTR as a primary metric for a regulated onboarding flow.