The Cohort Windows module finally explained why our seven-day view looked healthy while finance saw a mess. I still wish the spreadsheets came pre-built for AppsFlyer; we spent an evening mapping columns. Even so, we stopped arguing about which CAC was “real.”
Reviews
Not all glowing. All specific.
These notes refer to named modules on this site. Length and tone vary on purpose. One reservation appears because a classroom that never admits friction is selling something else.
★★★★☆
Incrementality Design Lab, March cohort. Clear on geo-split limits. Took a star off because the sample-size homework assumed we had more regional volume than we do. Would still send the media lead again.
Short version: Ledger Review hurt. Useful hurt. We had been blending brand search into paid social CAC and calling it blended efficiency. The written “claims to stop” page is now taped near the kettle.
Client in travel. We joined Measurement Residency after a festival spike destroyed our trailing CAC. The pairing sessions kept us from launching a holdout we could not power. No poetry here — just fewer embarrassing slides in the Monday meeting.
I came for Creative-to-Cost Mapping after our TikTok CPA looked stable while contribution fell. The fatigue lecture was the piece I replayed. Mirror Flowcore is slower than a vendor demo. That is the point, and also why my CMO needed convincing before we paid the Studio Seat fee.
Case study 01
Beauty DTC, Bangkok — window before creative
The team arrived reporting a 7-day CAC that finance never booked. After CAC Measurement Studio they published a 21-day observed window and a 45-day modelled tail. Reported cost rose by roughly a fifth in the first month. Contribution conversations shortened because both rooms used the same lag assumption.
They did not “fix” creative in class. They stopped treating a dying prospecting ad as an efficiency win. Six weeks later a small geo holdout (Incrementality Design Lab homework) showed that one prospecting set was mostly harvesting demand already in Line chats.
Case study 02
Regional marketplace — join keys before dashboards
A marketplace growth pod sent two people to Studio Seat and later expanded to a Residency. Their blocker was not theory: campaign IDs in the ad accounts did not match the warehouse. The first residency week was almost entirely naming. By week five they could produce a weekly CAC that internal audit would look at without flinching.
Limitation they still cite: Mirror Flowcore did not staff an engineer. Their data team did the pipeline work. We consider that a correct outcome.