Activation rate dropped from 41% to 32% over the last week and nobody knows why. Dig into the analytics and logs, find where in the funnel it broke and what changed, then draft an action plan with owners.

I'll start in Amplitude to find which funnel step lost conversion, then correlate the timing with deploys, Datadog error rates, and Sentry to pin the cause.

Breaking down the funnel2 steps

Segment the drop by platform

The break is the "First project" step on mobile web. Checking what shipped and what the logs say.

Confirming the cause3 steps

Open Datadog dashboard for /onboarding/new-project

Open the suspect deploy PR

Cause confirmed. Building a metrics view and an action plan with owners.

Found it. Activation dropped 41% → 32% and the loss is entirely at the Verify email → First project step on mobile web (−17%). The change is a clean step-down at 2026-05-31 13:52 UTC, which is exactly when PR #2904 ("Lazy-load project template picker") deployed. On mobile Safari, clients on cached HTML 404 on the new lazy chunk and throw TypeError: cannot read 'templates' of undefined - 9,841 Sentry events, Datadog client error rate up from 0.4% to 11.2%. I've opened the metrics view and an action plan with owners in Libra Computer.

Create incident issues in Linear

One P1 to revert + fix the lazy-load, plus two follow-ups for monitoring guardrails.

Draft · not sent

⠇ThinkingESC to interrupt

Waiting for LLM...·0ms·- tok/s·1/45

12 / 12