Reviews

What changed after the lab — and what still itches.

These notes refer to real Model Servicehub programmes. We keep the mild complaints in public on purpose. A studio that only publishes glow is not a studio you should trust with your Monday numbers.

Audience Signal Lab made our beauty show stop celebrating a spike that never returned. We now brief the editor from a 14-day curve. I still think week three assumes too much spreadsheet comfort, and I said so in the room.

Ploy S. · showrunner, Bangkok

★★★★★

House Partnership gave our in-house team a shared metric kitchen. The monthly pulse is dull in the best way. Nobody performs surprise anymore.

Niran K., head of creator ops · regional retailer

I only took Weekly Pulse Rituals. Forty-five minutes, three charts, one owner. My reservation: if your app export is messy, the ritual will expose that faster than it will fix it.

Fern · Chiang Mai

★★★★☆

Retention Atlas helped me recut openings without guessing. Faculty refused to promise a TikTok Shop connector that would last a year. I wanted the connector anyway. They were right.

Client in football highlights media

Desk Pass was enough. I could not travel to Phasi Charoen every week. The two written faculty replies were slower than I hoped — four days once — but the capstone checklist was specific to my podcast, not a template dump.

Arun, independent host

★★★☆☆

Useful on interpretation, thin if you arrived expecting SQL. Model Servicehub is a cutting-room for Creator App Analytics, not a data engineering school. I transferred my remaining curiosity to a developer friend, which the faculty actually encouraged.

Mina T. · product analyst sitting in on a creator team

Two colleagues reviewing documents at a desk

Case note 01

A Bangkok beauty show that was “up” and still shrinking

The team arrived with a screenshot of a million-view week and a quiet panic: returning viewers over 30 days had slipped for three months. Inside Audience Signal Lab they built a two-line wall chart — views versus 14-day return — and stopped letting the first line excuse the second.

After the capstone they changed thumbnail tests from “which image pops” to “which image keeps the same cohort for a second episode.” Shop revenue did not jump in week one. The Monday meeting did get shorter. Six weeks after the lab, their pulse owner reported that return had flattened rather than fallen, which was the actual goal.

Told with permission. Brand name withheld at the client’s request.

Case note 02

A football clip channel that recut the first eight seconds

Retention Atlas students from a highlights desk in Nonthaburi brought exports that looked healthy until the opening hook. Viewers left between second six and second eleven, then the remaining audience stayed. The “fix” they wanted was a louder watermark. The faculty asked them to recut the opening with the score already visible.

The following week’s curve still dipped — less steeply. They did not call it a win in public. They did add the recut to their experiment log, which is the habit we care about more than a single spike.

Browse programmes