“Cohort Intelligence forced us to stop averaging new and returning users on one chart. The first week felt pedantic. By week six our growth meeting had a smaller agenda and a truer one.”
Voices
What people said after the room closed.
We keep a 4.2 average from 67 written notes. That is not a podium score. It includes people who wanted more SQL than we teach, and people who wished Retention Mapping Studio had a second week. Both views belong here.
★★★★☆
“Attribution Without Guesswork did not give us a single model to worship. It gave us a list of claims we are no longer allowed to make in paid social. Finance liked that more than marketing did.”
“I am Tomas in Glasgow. The identity module in Signal Atelier stopped a merge that would have counted my partner and me as one power user. The recordings, though, are too quiet on laptop speakers.”
“A client in grocery delivery. We wanted Funnel Diagnostics to be a template we could paste. It is not. It is a method for sitting with drop-off that looks like a bug and might be a copy problem.”
★★★☆☆
“Event Taxonomy Architecture is the reason our warehouse is calmer. I still think nine weeks of Cohort Atelier is long if you only wanted a naming workshop. Ask for Signal Desk if that is you.”
Study 01
A wallet app that had three definitions of “activated”
A London fintech arrived with a proud activation rate and a support queue full of people who had “activated” without funding an account. During Cohort Intelligence we drew three cohorts: opened the app, passed KYC, and moved money. Only the third predicted week-four return.
The team did not add a new dashboard. They removed two events that were being used as success. Six months later their board pack showed a lower activation number and a higher funded-retention number. The visiting operator in the final fortnight asked only one question: “Which of these can you still explain if Mixpanel is down?”
Study 02
A fitness studio app that worshipped D1
The studio’s growth lead treated day-one return as a personality of the product. Retention Mapping Studio made them annotate the curve by class type. Indoor cycling returned; on-demand yoga did not, because the event fired on video start rather than a completed session.
They left with a narrower event and a less flattering D1. One tutor note still on file: “Do not celebrate a play-head. Celebrate a sweaty person who came back on Thursday.” They rated the programme 4 of 5; the missing star was for wanting more instructor-level analysis than we offer.
