Case #026: The campaign that only made money before noon

Case #026: The campaign that only made money before noon
An app-install offer, $100/day, running 24 hours a day because why wouldn't it. Aggregate ROI sat at a frustrating -8%. Right on the edge. I almost cut it as a near-miss.

Before killing it I pulled the hourly report — something I'd skipped at launch. The story was violent. From 6am to 1pm the campaign ran at +40% ROI. From 7pm to 2am it bled at -55%. The late-night traffic was cheap, plentiful, and worthless — a different user in a different mood, and the algorithm loved buying it because it was abundant. My profitable mornings were funding my catastrophic nights, and the blend hid both.

The fix: I now never assume a flat 24-hour day. Every campaign gets an hourly ROI breakdown by day 3, and I cut the hours that lose money rather than carrying them on the back of the hours that win. Day-parting isn't optimization, it's triage.

I shut off everything after 4pm and concentrated budget into the profitable window. $1,550 spent, $2,200 back. 42% ROI — built entirely out of hours that had been there the whole time, suffocated by the ones I refused to look at.

— The lesson: a near-breakeven campaign is often two campaigns in a coat — a winning daytime and a losing night — and the report that saves you is the one sorted by hour.