The situation
A residential cleaning company was running two lead sources side by side: Google’s Local Services Ads, which they managed themselves, and a Performance Max campaign. Both produced leads. Neither was being judged separately, so budget decisions were being made on a blended number that flattered the weaker channel.
What was really going on
Reporting that mixes a self-managed channel with a managed one hides the trade-off. The company could see total spend and total leads, but not which dollar was working harder — and so the temptation every month was simply to spend more across both.
What I did
- Split reporting so each channel was measured on its own cost per lead, and excluded the self-managed Local Services Ads spend from the managed performance picture so the numbers meant something.
- Rebuilt the Performance Max campaign’s inputs and tightened what it was optimizing toward.
- Tracked results weekly rather than monthly, which is what made a genuine week-over-week shift visible fast enough to act on.
- Reduced overall spend deliberately while the efficient channel was still scaling, instead of adding budget to protect volume.
Results
| Metric | Result | Context |
|---|---|---|
| Performance Max cost per lead | $78.70 → $16.32 | a 79% reduction |
| Performance Max leads | 2 → 7 per week | on lower total spend |
| Blended cost per lead | down 35% | best result in the reporting period |
| Total weekly spend | down 30% | with lead volume up 8% |
The takeaway
More leads for less money is almost never a clever bidding trick. It’s usually the result of measuring two channels separately for the first time, then moving money toward the one that was already working. You cannot make that decision on a blended number.
Want this looked at in your account?
Most of what I find in a new account is not exotic. It is a tracking setup nobody checked, a blended number hiding two very different channels, or budget going to searches that were never going to call. Request a marketing audit and I will tell you which of those you are dealing with.