Cost per lead is the most used metric in paid social and the most misleading when used alone. It measures the cost of an event, not of a result.
The arithmetic that changes decisions
Two creatives, same spend, different lead quality:
| Creative | Spend | Leads | CPL | Qualified rate | Qualified leads | CPQL |
|---|---|---|---|---|---|---|
| Ad A | €600 | 100 | €6.00 | 10% | 10 | €60 |
| Ad B | €600 | 50 | €12.00 | 40% | 20 | €30 |
On CPL, Ad A wins by a factor of two. On qualified leads, Ad B is twice as good. If you optimise on CPL, you will scale Ad A and cut Ad B — the exact opposite of the right decision.
Run your own numbers through the CPL vs CPQL calculator before you reorder the cut list.
This is not a rare edge case. Creatives that promise too much, or that target the broadest possible audience, produce cheap leads that do not qualify. Cheap leads are frequently the symptom.
Where qualification data lives
Qualification happens after the lead leaves the ad system: a call, a form review, a meeting. That means the data lives in your CRM or call log, not in the platform.
The bridge is lead routing: the lead leaves with its ad_id and landing page attached, so the destination system can store the source alongside the outcome. Later, the outcome comes back as a qualification flag on a lead whose source is known.
If routing and attribution live in different systems, this join is impossible and CPQL stays unmeasurable — which is why the two features are in the same product.
What “qualified” should mean
Pick one definition and keep it. The usual candidates, in increasing strictness:
- Contactable — the phone number or email works.
- Contacted — a human answered.
- Criteria met — the lead fits your target (budget, geography, need).
- Booked — an appointment exists.
- Showed — the appointment happened.
Stricter definitions are better signals and produce smaller numbers. The mistake is switching definitions mid-campaign, or letting two people on the team apply different ones.
The routine
- Route every lead out with its
ad_idand page (see the matrix routine). - Have the destination system record the qualification outcome against that source.
- Weekly, compute CPQL per creative × page, not just per creative.
- Reorder the cut list: cut on CPQL, using CPL only to explain why a combination is expensive.
- Keep both numbers visible. CPL tells you about the top of the funnel; CPQL tells you whether it is worth having.
When CPL is still the right metric
CPQL needs a sales process that produces data, and volume to produce it at a useful rate. If you are at 20 leads a week and nobody calls them, CPQL will be noise.
In that situation:
- Track CPL per creative, but treat it as provisional.
- Add the cheapest possible quality signal — did the lead answer, yes or no.
- Do not pretend the cheap creative is the good one. Say “we do not know yet” instead, and fix the data path before scaling.
Related
- How to read the attribution matrix — the table CPQL reorders
- Lead routing — how source and outcome meet
- CPL by creative — producing the numerator’s counterpart
- CPL vs CPQL calculator — run the two numbers on your own campaigns