
About the author
Diego Morales
Consultor de testing de push notifications (ex-iAdvize)
Diego Morales pasó seis años en iAdvize (2018-2024) como especialista senior en push notifications, ejecutando programas de testing de creativos para advertisers tier-1 en iGaming, finanzas, y utilities en EU y LATAM. Corrió 4,322 tests A/B/n. Se fue para publicar los datos que nadie más publica.
Background
Diego joined iAdvize in early 2018 as a junior creative analyst for the push-notification book. The job was meant to be running campaign briefs through the in-house creative engine and reporting CTR back to the AMs. Within eight months he was the one designing the test framework, because nobody else in the office had bothered to ask whether the "winning" creative had been declared a winner with n=180 or with n=8,000. The answer, depressingly often, was the first one.
What he learned over the next six years isn't in any deck iAdvize would let him publish. It's in the variance — the creative that "won" at n=200 and "lost" at n=2,000, the GEO where Chrome desktop converts at triple Chrome mobile despite identical creative, the day-7 audience fatigue cliff that nobody graphs because nobody runs tests that long. Between 2019 and 2024 he ran 4,322 A/B/n tests on push-notification ads. The database is his. He still references it weekly.
He left in March 2024 with no plan more elaborate than "publish the tables, properly this time." The writing started as internal memos to former colleagues, became a series of long-form posts shared on private Telegram channels, and is now this site. He still consults for advertisers who need a second opinion on their stat-sig conclusions. He still runs his own push tests, mostly to keep the database current, partly because stopping would feel like lying about being a practitioner.
What Diego writes about
- 01 Push-notification creative testing — six years and 4,322 tests of inside data
- 02 Push CTR/CR variance by GEO + browser + device — segmented panel data from his own tests
- 03 Audience fatigue curves for push — multi-week test designs that most agencies skip
- 04 Push frequency cap optimisation — including the "marginal-CR-vs-scale" tradeoff floor
- 05 Statistical significance applied to ad creatives — sample-size tables, CI design, power analysis
- 06 What he avoids: popunder (Marco's beat), native (not enough test data), CTV (post-2024, after his time)