Short answer
Click-based tools cannot see offline media, cannot tell a sale your ads caused from one that was coming anyway, and let each platform grade its own homework. On top of that, privacy rules now censor the online data they do see. Those are limits of what a click stream contains, not bugs. We model the whole business and check it with match-market tests.
Click data is useful for some things — campaign diagnostics, comparing creative, spotting a broken funnel. What it cannot do is tell you where the next dollar of a multi-channel budget should go.
A click-based tool sees 4 of these 10 channels. The other 6 report nothing — so they get cut first.
It cannot see most of what moves your business
Your TV. Connected TV. Radio. Billboards. Print — which still performs better than most people expect across the clients we have modeled. Word of mouth. None of it produces a click.
And the gap is not random. Offline often creates the demand that online then harvests. So the omission always tilts the same way — toward the bottom of the funnel.
It cannot tell caused sales from sales you had anyway
- The counterfactual problem
- Attribution records that someone touched an ad before buying. It cannot know whether they would have bought without it. Only the second question matters for a budget.
This is why branded search and retargeting report so well, and why both are usually over-funded. Both catch people who already decided.
Every platform grades its own homework
- Different lookback windows mean the same purchase gets claimed several times.
- View-through rules are set by the party with an interest in the answer.
- No platform can see the others, so none can settle the overlap.
- iOS 14 and GDPR censor a growing share of what these tools can see at all.
What we use instead
| Method | What it answers |
|---|---|
| Marketing mix model | What every channel contributes, offline included, at every spend level |
| A real baseline | What would have sold with zero ads |
| Match-market tests | Whether the model is right, measured for real |
| Testing on hidden months | Whether the model learned or just memorized |
“The question is not which touchpoints came before the sale. It is what would have happened if you had not run the ad.”
Is multi-touch attribution better than last-click?
It spreads credit more evenly among touchpoints it can see. It does not fix offline blindness or the counterfactual problem.
What about server-side tracking?
It recovers signal lost to browser restrictions. It does not make tracked media causal or untracked media visible.
Should we drop our attribution platform?
No. Keep it for in-channel diagnostics. Stop using it to set cross-channel budget.
Does testing replace attribution?
For budget decisions, mostly yes. For day-to-day monitoring, attribution is still more practical.
Find your wasted third
Want this run on your own numbers?
The two-week audit shows you where each of your channels sits on its own curve — what to cut, what to grow, and how much budget is sitting in the wrong place.
Book the two-week auditKeep reading