The setup. Push-to-sweepstakes on iOS, US tier-1. Numbers illustrative but the shape is real.
The move. Day 1-3 the tracker showed 0.38 ROI. Every instinct said cut it. I'd killed three campaigns that quarter on exactly this signal.
The numbers. Spend $2,100 across three days. Tracker revenue $800. But the network postback was running 48-72h behind because the advertiser batched conversions. By day 6 the same cohort had paid out $3,150 — 1.5 ROI on cohorts I almost murdered on day 2.
The lesson. Post-iOS, conversion timing got lumpy. If your offer has a delayed payout model — trials, app installs with in-app events, anything with a CRM step — your day-1 ROAS is fiction. You're not reading performance, you're reading reporting latency.
What I'd do differently. Build a cohort curve before scaling, not after. Track what percent of a day's revenue lands by hour 24 vs hour 72. Mine was 41% by 24h. Now I judge new offers against that curve instead of a flat break-even line, and I haven't knee-jerk-killed a winner since.
Arb Files
@ArbFiles
The week iOS killed my dashboard, not my campaign
This post was published in the Telegram channel Arb Files (@ArbFiles). All its posts are on the channel page.
Open on Telegram