Reviews

Notes from desks that had to defend a score.

These are enrolment reviews, not ad testimonials. Several mention a specific module. One is deliberately mixed.

After Scoring Studio I could show why our “efficient” incentivized network was poisoning Day-7. The decay module is the piece I still replay.

Pimchanok Saetang, UA lead at a Bangkok fintech

Arthit here, Chiang Mai. Tiny team. The store-listing-as-channel session stopped us dumping brand search into “organic leftover.”

Arthit · Chiang Mai

★★★★☆

Verified enrolment. The ATT consent session finally gave us a way to stop treating iOS as a single blob. Would have liked more Android SKAd-adjacent examples.

Platform review · April 2026

Homework during Calibration Lab weeks ran heavier than advertised, and the templates assume SKAN postbacks are already joined cleanly. We still use the rubric every Monday.

A performance marketer at a Thai gaming publisher

I came in sceptical of “quality scores” after two agencies sold us composite indices nobody could explain. Mali’s critique nights were blunt. We left with weights written in plain Thai and English so product and finance could argue in the same document.

Krittika N., product operations · Bangkok

Case notes

Two desks, two different scores

Team reviewing documents around a table

Fintech UA desk, Bangkok — retired two sources, kept spend flat

A consumer lending app arrived with four paid networks and a blended ROAS that looked respectable. Scoring Studio’s week-three decay exercise showed Network B’s Day-7 retained users falling to 11% while CPI stayed the “winner.” They cut B and a thin incentivized source, redistributed budget, and recorded a 2.1× lift in Day-7 retained users at the same weekly spend over the following six weeks. The limitation: they still cannot score web-to-app traffic cleanly because MMP web attribution remains incomplete.

Colleagues collaborating with laptops in an open office

Hypercasual publisher — organic halo finally sat on the ledger

The growth pair treated store listing conversion as “just ASO.” After the organic baseline calibration module, they placed listing, brand search, and two paid networks on one scorecard. Paid still looked cheaper on CPI; listing scored higher on 14-day retained value. They stopped pausing listing experiments during paid spikes. Revenue per retained user rose modestly (about 17% over a quarter). They did not get a miracle. They got a score they could defend when the publisher asked why CPI went up on purpose.