Open any agency report and cost per lead is on the first page, usually large, usually with a downward arrow next to it. Underneath: impressions, reach, click-through, maybe a chart of form fills by week. Fourteen slides in and nobody has said how much money the client banked.
The problem with cost per lead isn't that it's meaningless. It's that it's trivially easy to improve without improving anything you care about, which makes it worse than meaningless when it's the number your supplier is being judged on.
How to halve your cost per lead by Friday
Four moves, none of them dishonest, all of them available to any agency under pressure at the end of a month:
- Widen the targeting. Drop the job-title filter, add the adjacent industries, extend the geography. More clicks, cheaper clicks, more forms.
- Cut a field off the form. Every question you remove lifts completion rate. "What's your current turnover" is usually the first to go, and it was the only qualifying question you had.
- Change the offer to something easier to say yes to. "Book a call" becomes "download the guide". A guide download costs a fraction of what a booked call costs, because it asks for a fraction of the commitment.
- Count more things as leads. The newsletter signups, the webinar registrants, the person who chatted to the widget and left an email.
Suppose you were at £120 a lead across 50 leads a month. Run all four and you'd plausibly land somewhere near this:
| Before | After | |
|---|---|---|
| Spend | £6,000 | £6,000 |
| Leads | 50 | 150 |
| Cost per lead | £120 | £40 |
| Leads that fit the business | 18 | 22 |
| Booked calls | 12 | 9 |
| Deals closed | 3 | 2 |
That's a hypothetical, not a client result, but the shape of it is what happens. Cost per lead fell by two thirds. The agency had an excellent month. You have 128 more names to process, one fewer deal, and a salesperson who now believes the leads are rubbish.
Nobody lied. The number genuinely went down.
"Are the leads in my report real people?"
That question gets asked in enough procurement calls and enough Google searches to be worth answering directly, and the answer is: sometimes, partially, and the report is not designed to tell you.
A "lead" in an agency report is whatever the tracking was configured to count. That usually includes form submissions with a real email address, and it frequently also includes newsletter signups, gated-content downloads, chat sessions where somebody typed anything at all, and phone calls over thirty seconds. Every one of those is a real event. Not every one is a real person who wants to buy from you.
Ask for the raw export rather than the dashboard, pick twenty rows at random, and look them up. You'll find out in ten minutes what six months of reporting never told you. If the agency can't produce that export, the number on page one is describing their activity rather than your pipeline.
What about cost per qualified lead?
It's a genuine improvement on cost per lead and it's the metric most people reach for next, so it's worth saying where it still falls short.
Dividing spend by qualified leads rather than all leads removes the easiest gaming move, which is counting junk. It doesn't remove the harder one, because "qualified" is a definition your supplier can also loosen, quietly, over a quarter, without telling anybody. Unless the qualification criteria are written down, agreed by both sides and audited occasionally, cost per qualified lead decays into cost per lead with an extra word in front of it.
Use it if you like it. Just don't put it on page one on its own, and make sure somebody outside marketing owns the definition.
This has a name
Goodhart's law: when a measure becomes a target, it ceases to be a good measure. Charles Goodhart was writing about monetary policy in 1975, but it describes agency reporting better than anything written since.
The general form: any metric a supplier can improve without improving your bank balance will eventually be improved without improving your bank balance. That isn't cynicism about agencies, most of whom are trying hard. It's what happens structurally when you pay people for activity and measure them on a proxy.
Cost per lead is not the only offender, just the most common:
| Metric | How it's improved without helping you |
|---|---|
| Cost per lead | Ask for less commitment, from more people |
| Reach and impressions | Stop caring who sees it |
| Reply rate on outbound | Ask a smaller question |
| Pipeline value | Let salespeople estimate deal size on a Friday afternoon |
| MQLs | Change the definition of an MQL |
| Cost per click | Bid on cheaper, less relevant terms |
There's a useful outside data point on the CPL question specifically. Metadata's 2026 B2B advertising benchmarks, drawn from 153 B2B advertisers and $57.6m of spend producing 211,000 leads in 2025, put LinkedIn at $202 a lead and Google Ads at $524. If cheap leads were the goal, every B2B company in that dataset should abandon Google tomorrow. They don't, because the two channels catch people at different points and a Google search lead is frequently worth several LinkedIn form fills. Their own conclusion is the sensible one: a cheaper lead is not automatically a better one.
The three numbers
We've never needed more than three to tell whether a business is growing or just getting busier.
1. Qualified opportunities per week
Not leads. An opportunity is a named person, at a business that fits what you sell, who agreed to a conversation about buying and then turned up to it. The "turned up" clause is doing real work: no-shows are a marketing problem masquerading as a diary problem, and counting them flatters everybody.
Weekly, not monthly. A monthly number hides a dead fortnight, and you'll only notice in the quarter after the one where it mattered.
Write down the number that would be enough before you measure, otherwise whatever you find will feel acceptable. For a founder-led B2B business turning over £10k to £60k a month, the honest requirement is usually somewhere between three and eight a week. Most are doing two.
2. Median time to first human contact
Median, not mean. One enquiry answered after four days drags a mean into fiction while ten answered in six minutes look fine underneath it.
Measure from the timestamp on the enquiry to the timestamp on the first message or call from a person who knows something about your business. An automated acknowledgement doesn't count and shouldn't; it's useful, it buys you patience, and it is not contact.
This is the cheapest number in the business to move and the most likely to be quietly awful, because it belongs to nobody. The 2011 Harvard Business Review audit of 2,241 US companies found 23% never responded to a web enquiry at all, and the average among those that did was 42 hours. Nothing we've seen since suggests that got dramatically better, and everything about buyer expectations suggests the tolerance for it got worse.
3. Cash collected per opportunity created
Not revenue. Not invoiced. Collected, in the bank, from the opportunities that entered the pipeline in a given month, measured for as long as it takes them to close.
This is the number that ends arguments, because it prices your leads properly rather than pricing your marketing activity. Twenty opportunities that pay you nothing for four months are worth less than eight that pay in three weeks, and no dashboard built around cost per lead will ever tell you that.
It's also the only one of the three your accountant can check.
Why three, and why they hold each other honest
The three can't be gamed against each other, which is the entire point of using them as a set rather than picking a favourite.
| If you | Number 1 | Number 3 | What you learn |
|---|---|---|---|
| Loosen targeting for volume | Up | Down | The extra volume was junk, immediately visible |
| Only chase perfect-fit accounts | Down | Up | Great deals, not enough of them |
| Speed up follow-up | Up | Up or flat | The one move that improves both |
| Raise prices | Down or flat | Up | Usually the right trade, now provable |
You're forced to hold volume and value at the same time, which is what running a business actually feels like and what a single headline metric always flattens.