Creative attribution is the practice of measuring cost per lead for each individual creative and each individual landing page, instead of for the campaign that contains them.
The definition sounds like a detail of reporting. It is not: it changes which decisions are available to you.
The three levels of measurement
| Level | The number you get | The decision it enables |
|---|---|---|
| Account | blended CPL | increase or decrease total budget |
| Campaign | average CPL | shift budget between campaigns |
| Creative × landing page | CPL per combination | cut the losing combination, keep the creative |
Only the third level tells you why a campaign is expensive. At the campaign level, a €7.91 CPL and a €7.91 CPL can be completely different situations: one campaign has four creatives that all perform the same, another has one excellent creative carrying three bad ones.
Why the campaign level fails
Campaign-level reporting is an average, and averages hide the variance that decisions need.
Take four combinations inside one campaign:
| Combination | Spend | Leads | CPL |
|---|---|---|---|
| Ad A × LP 1 | €420 | 100 | €4.20 |
| Ad A × LP 2 | €640 | 50 | €12.80 |
| Ad B × LP 1 | €300 | 40 | €7.50 |
| Ad B × LP 2 | €380 | 30 | €12.67 |
| Campaign | €1,740 | 220 | €7.91 |
The action that follows from the campaign number (“CPL is €7.91, it is acceptable”) is to do nothing. The action that follows from the matrix is to kill Ad A × LP 2, investigate what LP 2 does worse, and increase budget on Ad A × LP 1.
Same data. Different level. Different decision.
What makes it hard to do
Measuring at the creative level requires three things that most stacks do not have together:
- The click must survive. If the identifier is lost between the ad and the form, the lead cannot be tied to the ad. Browser restrictions make this the hard part — see tracking after iOS.
- Spend must be known per ad, not per campaign. That comes from the ad platform, per
ad_id. - The mapping between a creative and its
ad_idmust be reliable. When ads are built by hand, this mapping is a convention that drifts. When the tool creates the ad, it is a fact in the database.
The operational consequence
Creative attribution turns one question into three:
- Which creative should I cut? The one whose combinations all lose.
- Which landing page should I cut? The one that loses across several creatives.
- Which combination should I scale? The one that is cheap at volume.
Without the combination, these three collapse into “the campaign is doing badly”, which is not actionable.
Related
- CPL by creative — how to measure it step by step
- Real CPL vs platform CPL — which number to trust
- How to read the attribution matrix — turning the table into actions