The Signal Chamber forced us to throw out “content_viewed” as a north star. Painful in the room, quieter dashboards after.
What the circles actually changed
These notes refer to named rooms on this site. We keep the mild complaints; a wall of praise is not a reading list.
Week three of the Lab was denser than advertised. I kept up, but I would not send a junior PM alone. The taxonomy letter we left with is still on our wiki.
★★★★☆
Solid on windows. Thin on Android backgrounding. I bought the Mobile Event Grammar weekend afterwards and that closed the gap.
We came for replay craft and were told, correctly, that we were watching too much. The sampling rules from the Atelier are now in our privacy review.
Case: a bill-pay app that trusted D1
A Cardiff bill-pay product arrived with a handsome day-one retention line and a board pack that treated it as habit. In the Cohort Signal Lab they rebuilt the entry event around “first successful payment,” then moved the window to fourteen days to match how households actually open the app.
The new curve looked worse for six weeks. Finance disliked the honesty and then, unusually, kept it. The guest’s own summary: the Chamber did not make the product more loved; it made the lie unusable.
Case: replay without a spectacle
A London fitness studio’s app had a library of session recordings no one had time to watch. The Session Replay Atelier cut the sample to failed checkout plus a 4% random slice, and wrote consent copy that named the vendor.
They did not “find a hidden insight” every week. They found one broken wallet sheet and a legal review that had been postponed. That was enough to justify the shorter room.
