Cost per lead is the number that hides the problem
A cheaper lead is not a cheaper customer. For most service businesses the two numbers move in opposite directions, and only one of them pays the wages.
The number everyone optimises
Cost per lead is easy to measure, updates daily, and appears in every platform dashboard by default. That combination makes it the number agencies report and clients ask about. It is also the number that can fall for months while your business gets worse.
The reason is simple: a lead is not a commitment. Lowering the barrier to enquiry, a shorter form, a vaguer offer, a broader audience, reliably produces more leads at a lower unit cost. It just as reliably produces leads with less intent behind them.
A falling cost per lead with flat revenue is not an improvement. It is a warning.
What to measure instead
Cost per customer is spend divided by customers won, over a window long enough for your sales cycle to complete. For an appointment-based service business that is usually 30 to 90 days. If your average job takes two weeks to close, a seven-day reporting window will systematically credit the wrong campaigns.
Alongside it, track payback period: how long it takes for a customer to return what you spent acquiring them. A customer worth A$1,200 up front is a different proposition from one worth A$1,200 spread over eighteen months, even though both look identical in a lifetime-value column.
How to actually calculate it
You need three things: spend by campaign, enquiries by campaign, and outcomes by enquiry. Most businesses have the first two and not the third, which is why the calculation never gets done.
The minimum viable version is a spreadsheet. Every enquiry gets a row with its source, the date, and whether it became a customer and for how much. Ninety days of that beats any attribution tool you have not configured properly, because it reflects what actually happened rather than what a pixel thinks happened.
Once that exists, the ranking almost always changes. In our experience the campaign with the cheapest leads is rarely the campaign producing the most profitable work, and the gap is often large enough to reallocate a third of the budget.
The trap in blended reporting
If you add up the conversions each platform claims and the total exceeds the enquiries that reached your inbox, you are double-counting. Two platforms both take credit for the same customer, so blended cost per customer looks better than reality and every optimisation decision inherits the error.
Reconcile one month, line by line, against your CRM or inbox. It is tedious and it is usually the highest-value hour you will spend on your marketing that quarter.
What good looks like
One number you trust, reported monthly, reconciled to revenue. Cost per customer by campaign, with payback period beside it, and a note on what the attribution cannot see. Everything else, impressions, clicks, cost per lead, engagement rate, is a diagnostic that helps you understand why that number moved. None of them are results.
If your current reporting cannot answer “what did a customer cost us last month, by channel”, that is not a reporting problem to solve later. It is the reason the other decisions are hard.