Case #033: The 72-hour campaign rescue

A gambling offer was sinking — day 5, $1,640 spent, $980 back, negative 40%. The choice every buyer faces: cut it, or commit 72 hours to a structured rescue. I gave it the 72 hours, run by a fixed diagnostic so I wasn't just flailing. Here's the turnaround playbook.

Hour 0 — isolate the bleed:

— Break performance by every dimension at once: source, geo, device, hour, creative. One sub-source was 47% of spend at 8% conversion. Paused it immediately. That single cut changed the trajectory.

Hour 0-24 — fix the biggest leak only:

— One change at a time, measured: I didn't touch creatives yet. Just killed the bad sub-source and watched 24 hours. ROI moved from -40% to -8%. The leak was the source, not the offer.

Hour 24-48 — address the second leak:

— Now the creative: frequency had hit 3.1 on the survivors. Swapped in two banked creatives. ROI crossed into positive, +14%.

Hour 48-72 — confirm and commit:

— Hold all changes, add nothing new: let it stabilize. It settled at +31%. Only then did I resume scaling on the ladder.

— The rule: never make two fixes in the same 24 hours, or you'll never know which one worked.

The rescue ran the campaign from -40% to a finish of $2,910 spent, $3,810 back, 31% ROI. A campaign I'd have buried on day 5.

The lesson: a sinking campaign rarely has one wound — diagnose every dimension at once, then fix them one at a time, because the cut that saves you is usually a single bad sub-source hiding in the average.