Before changing a bid or pausing a campaign on an account we take over, we run the same 10-point checklist. No magic: most plateaued accounts fail on 2 or 3 of these points, and they are rarely the ones the client suspects.
Structure: points 1 to 3
1. Consolidation: fewer campaigns, more budget per campaign. Ten campaigns at $50/day learn slower than one at $500. Fragmentation is the most common mistake we see.
2. Learning phase exits: if most of your ad sets are stuck in limited learning, your structure is preventing the algorithm from converging.
3. Exclusions and overlap: audiences cannibalizing each other across campaigns, existing customers not excluded from acquisition.
“Most plateaued accounts fail on 2-3 specific points: fragmented structure, broken signal or creative fatigue.”
Signal: points 4 to 6
4. CAPI active and deduplicated (shared event_id between pixel and server). We wrote a full article on this because it is broken that often.
5. EMQ at 6+ on the Purchase event: below that, Meta reconciles your conversions with clicks poorly.
6. Events aligned with your real funnel: optimizing on Purchase with 10 purchases a week asks the algorithm to aim at a target it almost never sees. Sometimes the intermediate event is the right temporary choice.
Creative: points 7 and 8
7. Volume and freshness: how many new concepts enter the account each month? Below 5, creative fatigue alone often explains rising CPAs.
8. Angle diversity: if every ad sells the same benefit to the same segment, you are paying more and more for the same audience.
Measurement: points 9 and 10
9. MER tracked alongside platform ROAS: total revenue divided by total spend, week after week. The one metric that does not lie.
10. Consistent attribution windows: comparing a month on 7-day click against a month on 1-day view is apples and oranges, and the conclusions drawn from it are wrong.