Updated: 4 August 2026 · Digital marketing, Lead generation

There's a conversation I've had dozens of times. A client shows me their campaign dashboard, points at the cost per lead and says: "we've brought it down from €40 to €18." Then I ask how many of those leads actually bought, and the room goes quiet.

Lowering cost per lead is the easiest optimisation to achieve and the easiest one to sell. It's also, by some distance, the one I've watched lose the most money.

Two metrics that get confused constantly

Cost per lead (CPL) is simple: ad spend divided by the number of contacts you received. Spend €2,000, collect 100 form fills, your CPL is €20. The platform calculates it for you. You don't have to do anything to see it.

Cost per customer (CAC) is spend divided by the people who actually bought. If 4 of those 100 forms turn into signed clients, your cost per customer is €500.

The distinction isn't academic. CPL measures how many people raise their hand. CAC measures how many people pay. The gap between those two numbers is invisible to the ad platform and expensive to you.

Why the whole system pushes you toward CPL

It's no accident that most businesses optimise the wrong number. Three forces push in that direction:

The platform can only see as far as the form. Google and Meta know exactly when someone submits a form, because it happens inside their ecosystem. Everything after that — the call, the meeting, the proposal, the signature — happens in your CRM, on a salesperson's phone, or in a spreadsheet. If you never send that back, the algorithm optimises blind toward the only thing it can measure: volume of cheap form fills.

CPL moves immediately. You can cut it in a week by broadening targeting and stripping fields off the form. Cost per customer needs a full sales cycle to be trustworthy, and in high-ticket sectors that can be one to three months.

It reports beautifully. "We reduced cost per lead by 40%" reads brilliantly in a monthly report. "We increased cost per lead but we think the leads are better" requires trust and data almost nobody is collecting.

What this looks like in a business that sells to other businesses

Here's a worked example. The numbers are illustrative, but the shape is one I see repeatedly, particularly with companies selling professional services into the Spanish market.

A consultancy runs two campaigns on a €4,000 monthly budget.

The broad campaign targets a wide audience with a "download our free guide" offer. It generates 250 contacts at €16 each. Most of them wanted the guide, not the consultancy. Sales manages to reach 70, books 9 meetings, and closes 2 clients. Cost per customer: €2,000.

The narrow campaign targets a specific job title in companies above a certain size, and the ad states the engagement starts at €5,000. It generates 35 contacts at €114 each — a CPL seven times worse. But sales reaches 30 of them, books 18 meetings, and closes 5 clients. Cost per customer: €800.

By the metric on the dashboard, the second campaign looks like a failure. By the metric that pays salaries, it's two and a half times better.

I've watched profitable campaigns get switched off for exactly this reason.

Closing the loop without building a big system

You don't need expensive software to start. You need to connect the form to the sale. In order of effort:

1. Tag where every lead came from. UTM parameters plus a hidden field on the form is enough. Without this, nothing else is possible.

2. Track a status for every contact. Four states will do: new, contacted, qualified, customer. A simple CRM or a shared spreadsheet both work at low volume. The tool matters far less than somebody actually keeping it updated.

3. Feed the outcome back to the platform. This is the step that changes results. Both Google Ads and Meta accept offline conversion imports: you tell them "this specific lead became a customer," and the algorithm starts looking for people resembling buyers rather than people resembling form-fillers. That's the difference between optimising toward noise and optimising toward revenue.

4. Calculate CAC per campaign, not overall. The aggregate number hides the interesting part. There's almost always one campaign that looks expensive on CPL and brings the best clients, and a cheap one that closes nothing.

What changes in your campaigns

Once cost per customer becomes the target, decisions invert:

  • The form stops being an obstacle and becomes a filter. Adding qualifying questions lowers lead count and raises quality — covered in detail in how to qualify leads before they reach your sales team.
  • Stating your price stops being frightening. It scares off people who can't afford you, which is precisely what you want to happen before you spend sales time.
  • Targeting narrows. Forty people with real intent beat two hundred browsers.
  • Budget moves toward what's profitable rather than toward what's cheap.

That shift is the foundation of how I approach qualified lead generation for high-ticket work, and it's a different job from advertising management aimed purely at driving traffic.

When CPL genuinely is useful

I don't want to suggest CPL is worthless. It earns its place in two situations:

As a short-term alarm. If your CPL triples overnight, something broke — the landing page, the tracking, or a competitor arrived with a serious budget. It's an excellent fault detector.

When the ticket is small and the cycle is instant. If you sell a €30 product bought on impulse, lead and customer sit so close together that the distinction nearly disappears. The problem appears as the ticket rises and the cycle lengthens.

Frequently asked questions

How long before cost per customer is reliable? At least one full sales cycle, preferably two. In high-ticket sectors like finance or real estate that usually means one to three months. Before that you have an estimate, not a number.

What if my volume is too low to be statistically meaningful? That's the normal situation for a small business. With four or five clients a month you can't run statistics, but you can do something better: look at each closed client individually and trace where they came from. At low volume, case-by-case analysis beats any average.

Do I need a CRM? It helps a lot, but it isn't required to start. A spreadsheet with lead source, date and status gets you most of the value. A CRM becomes necessary as volume grows or when several people handle the same contacts.

My cost per customer came out enormous. Now what? First check the number is real — unattributed sales are extremely common. If it holds up, the problem sits in the campaign, the landing page or the sales process, and it's worth knowing which before touching budget.

Can I optimise both at once? Yes, with a clear hierarchy: cost per customer rules, cost per lead is a secondary indicator. When they conflict, the one measuring actual money wins.

If you're spending on acquisition and can't say what it costs you to close a client, that's the first number to put on the table. Tell me how you're generating leads today and I'll give you an honest read on where the room for improvement is.