Conversion Lab Notes
Conversion Lab Notes
@ConversionLabNotes

Myth: a winning variant helps all visitors. False — pooled wins routinely hide segment reversals (Simpson's pa

Myth: a winning variant helps all visitors. False — pooled wins routinely hide segment reversals (Simpson's paradox).
A +4% aggregate can be +12% on mobile and −6% on desktop. The blend looks like a win; half your traffic lost.
Pre-declare 2–3 segments (device, new vs returning, traffic source). Don't slice 20 post-hoc — that's just peeking by another name.
Read the number, not the story. [Simpson's paradox · 2–3 pre-set segments]
This post was published in the Telegram channel Conversion Lab Notes (@ConversionLabNotes). All its posts are on the channel page.
Open on Telegram

Latest posts in “Tools and infrastructure”

All channels in this category →

реклама в сети

Готовы запустить рекламу через сеть public.tg?

Новый оффер, продукт, GEO, кейс, событие или партнёрский запуск — соберём маршрут под задачу и отдадим медиаплан.

Telegram для медиаплана: @AFFtop_connect. Быстрый тест: $20 за канал, $1000 за пакет по сети.