A creative-axis change cut cost per lead 32%.
Two months later, it was worse than before the change.
Media- and campaign-level reports showed nothing unusual. Budget and targets had held for months; only cost per lead was drifting upward.
The month before each intervention is indexed to 100. Budget, clicks, conversions and cost per lead are all measured against that month.
Splitting by creative rather than by channel showed the gap: 41% of conversions came from one creative axis built around a specific occupation group, at 44% of the account's average cost per lead. Most of the budget sat outside that axis.
We did not raise budget. We scaled the high-efficiency axis and cut the low-efficiency one. The next two rounds were an offer-copy swap, then a network cleanup on the search campaign.
The three interventions happened at different times. Each row compares the month before that intervention to the month after.
| Intervention | Budget | Clicks | Conversions | Cost / lead |
|---|---|---|---|---|
| Month before (baseline) | 100 | 100 | 100 | 100 |
| 1 · Creative-axis swap | 102 | 122 | 149 | 68 |
| 2 · Offer-copy swap | 96 | 98 | 160 | 60 |
| 3 · Search network cleanup | 101 | 134 | 108 | 94 |
| Two months after #1 | 123 | 76 | 69 | 178 |
That last row is the point of this case. Running the same creative axis two more months cost 24% of the clicks and pushed cost per lead 78% above where it started — on 23% more budget. A single improvement does not hold. Season and competitive pressure are mixed in too, so not every change in the table can be credited to the intervention.
The shift was from finding one winning axis to measuring how fast an axis wears out. We keep a backup axis running rather than riding a winner to the end, and we time the swap off the wear signal instead of waiting for the monthly report. The cases below came from the same discipline.
