Why does my cost per lead go up when I increase the budget?
Because you buy the cheapest demand first. Raising budget means entering auctions you previously skipped — broader queries, worse positions, weaker hours — and those clicks convert less well. This is normal market behavior, not a broken account. The real question is whether the marginal lead still books at a rate worth paying for, and blended cost per lead cannot answer that.
Average cost hides the decision
The number on your dashboard is an average across all leads, including the cheap ones you would have gotten at any budget. When you scale, the average moves slowly while the cost of the next lead moves quickly. You are making decisions with the wrong number.
The number you want is marginal cost per lead: the change in spend divided by the change in leads between two periods. Written out, that is (spend after − spend before) ÷ (leads after − leads before). It is blunt, it needs periods of comparable demand, and it is far more honest than the average.
What the ratio tells you
Suppose spend rises by a fifth and leads rise by a twentieth. The marginal cost of those extra leads is roughly four times your average cost per lead. That is the price of expansion, and it may still be worth paying if those leads book and the jobs are large.
If spend rises and leads do not move at all, you have hit a ceiling — usually available demand in your service area, not a bidding problem. More budget will not manufacture homeowners with broken equipment.
Legitimate increase versus symptom
- Legitimate. Impression share lost to budget was high before the increase and fell after it. You bought share you were previously priced out of.
- Symptom. Impression share barely moved but spend rose, which means the extra money went to broader matching or looser geography rather than more of the same demand.
- Symptom. Conversion rate fell sharply rather than gradually, suggesting the traffic changed character rather than thinned at the margin.
- Neither. A competitor entered the auction and everything got more expensive regardless of what you did.
The decision rule that actually applies
Compare marginal cost per booked job to the contribution a job leaves after the cost of doing it. Not cost per lead against a target you set last year. If the marginal job still contributes, keep buying; when it stops, stop.
Doing that requires booking rate and job value by campaign, which means the ad platform has to be joined to the job record. That join is the substance of revenue intelligence, and it is why business intelligence built on operational data answers scaling questions that the ad platform structurally cannot.
Topics: cost per lead · marginal cost · budget scaling · unit economics
Have a version of this question about your own business?
The useful answer usually depends on which systems you run and how they're connected. That's a conversation, not a blog post.