Your cost per lead is lying to you
The metric most B2B teams optimise hardest is the one most likely to be actively destroying their pipeline. Here is the mechanism, and how to check in an afternoon.
Every few months we meet a company with an enviable cost per lead and a sales team that has quietly stopped trusting marketing. These two facts are not a coincidence. One causes the other.
The mechanism is simple enough that it is embarrassing, and it survives because the number it produces looks like success.
What you told the algorithm
Google's bidding is a machine that finds you more of whatever you called a conversion. It is extremely good at this and it has no opinion about whether your definition was sensible.
If 'Lead' fires on any form submission — a demo request, a newsletter box, a gated PDF, the contact form in the footer — then you have told the machine that a newsletter signup and a demo request are the same thing. It will buy you whichever is cheaper. It will buy you a great deal of it, and your cost per lead will fall, and you will be pleased.
You have optimised for the cheapest possible human with an email address. That is exactly what you asked for.
How to check in an afternoon
You do not need a project for this. Export last quarter's closed-won deals from your CRM. Export the same period's conversions from your ad platform. Then answer one question: what share of closed revenue came from each campaign?
Now compare that ranking to the one your platform dashboard shows. In our experience the two orderings frequently disagree, and sometimes they invert entirely — the campaign with the best cost per lead turns out to have produced no revenue at all.
- If the rankings match, your conversion definition is fine and you can stop reading.
- If they disagree, the number you have been optimising is not measuring the thing you care about.
- If you cannot produce the export at all, that is the finding. Fix the plumbing first.
The fix, and why it is unpopular
Pipe closed-won revenue back into the ad platform as an offline conversion, and let it optimise toward that instead. Within about six weeks the algorithm changes what it buys, because you have finally paid it for the right outcome.
Your cost per lead will get dramatically worse. For one client it went from €41 to €173, and the pipeline doubled while spend halved. That is not a paradox: you stopped buying four hundred worthless leads and started buying eleven good ones.
The hard part of this work is never technical. It is the three meetings where you explain why the number on the quarterly slide is supposed to go up.
The uncomfortable version
In the engagement above, the previous agency was hitting every KPI in their contract. They were not lazy or dishonest. The KPIs were simply wrong, and everyone — client included — had agreed to them.
That is the part worth sitting with. Nobody in that room was doing a bad job by the terms they had set. The terms were the problem, and no amount of diligence inside a bad definition rescues it.
Ran acquisition in-house at a Stockholm marketplace through the iOS 14 attribution collapse, which she describes as the most educational eighteen months of her career. Deeply suspicious of any ROAS figure she did not reconcile herself.