Attribution honest enough to
spend against.
Most attribution is theatre: a dashboard that reconciles to nothing and assigns credit by rule rather than cause. The models a finance team will actually fund measure incrementality, are agreed before the spend, and stay honest about what they cannot know.
A dashboard that convinces no one who pays
Most attribution is theatre. It produces a confident dashboard, a tidy split of channel credit and a story everyone in the marketing meeting can live with. It also reconciles to nothing the finance team recognises, which is why the spend conversation never gets any easier.
The problem is not that the numbers are wrong in some technical sense. Last-click, first-touch and the various weighted models all do exactly what they were built to do. The problem is that they assign credit by rule rather than by cause, and a rule cannot tell you what would have happened if you had not spent the money.
That distinction sounds academic until the budget review, when someone who controls the money asks a simple question — if we turned this channel off, what would we lose — and the dashboard has no answer. Everything it measures assumes the spend worked. None of it tests whether it did.
It flatters everyone in the room
Attribution this weak survives because it is comfortable. Last-click flatters the channels that sit closest to the sale, first-touch flatters the channels that sit furthest from it, and every team can find a model under which its own line looks essential.
So nobody in the room has an incentive to break it. The honest number is almost always smaller than the reported one, and it is rarely the person presenting who volunteers that. The model that makes everyone look good is the model that gets renewed.
Three habits keep it in place, and each of them feels reasonable on its own.
Credit is counted inside the platform that sold the media, so the channel grades its own homework and always passes.
Performance is measured against a target set in the same plan, so the plan can never be shown to have missed.
The number is never reconciled to revenue the finance team recognises, so the gap between reported and real is never named out loud.
Incrementality, not credit
A finance team does not need a prettier attribution model. It needs a number it can defend to the board, which means a number that answers one question honestly — how much of this outcome would not have happened without the spend.
That is incrementality, and it is a different discipline from attribution. Attribution divides a result that already exists. Incrementality measures the difference the spend actually made, usually by holding some of it back and watching what changes. It is less flattering and far more fundable.
A measurement model a CFO will sign off tends to share four properties, none of which are about the sophistication of the maths.
Measures incrementality, not credit
The question is what the spend changed, not which touch to reward.
Agreed before the money moves
The method and the success criteria are written down before the campaign, not chosen afterwards to fit the result.
Reconciles to the finance number
It ties back to revenue the business actually books, not a platform-reported conversion.
Honest about what it cannot know
It reports a range and its assumptions, so the confidence is earned rather than implied.
Agree the rules while you can still be wrong
The single most valuable thing you can do to make measurement fundable costs nothing and happens before any media is bought. Write down how you will judge the campaign, and get finance to agree to it, while the outcome is still unknown.
The reason is behavioural, not technical. Once the results are in, every choice of method becomes an argument about the result. Deciding the holdout, the window and the success threshold in advance removes the temptation to select, after the fact, whichever definition makes the quarter look best.
This is also the moment to agree what a disappointing result would look like and what you would do about it. A measurement plan that has no version of the answer where you spend less is not a measurement plan; it is a budget defence with charts.
A range you can act on
Honest measurement produces ranges, not decimal points. A geo holdout or a properly designed test tells you the effect sits somewhere between two numbers, with a level of confidence you can state plainly. That is less satisfying than a single figure and considerably more useful.
A range is enough to act on. If the worst credible case still clears the cost of the spend, you scale it. If the best credible case does not, you stop. The decision rarely needs three decimal places; it needs to know which side of the line the whole range sits on.
The teams that measure well are comfortable saying they do not know something precisely. The teams that measure badly report false precision, which holds up right until the first time reality contradicts the dashboard — and after that the credibility does not come back.
Three moves this quarter
None of this requires a new platform, a data science team or a rebuilt stack. It requires being willing to measure one thing honestly and to accept a smaller, truer number than the one you report today.
If you want measurement a finance team will fund before the next planning cycle, three moves are worth making now.
Pick one channel and run a holdout, so you have one number that describes cause rather than correlation.
Reconcile that channel's reported result to the revenue finance recognises, and name the gap out loud.
Agree the method and the success threshold for your next campaign before it launches — in writing, with the person who controls the budget.
This is how our Data & Decision Intelligence work is built: measure incrementality rather than credit, agree the rules before the spend, and report a number honest enough that the person funding it can defend it upward.
Explore Data & IntelligenceLeads our data and measurement practice across India, Dubai and the US. Spends most of their time turning marketing dashboards into numbers a finance team will sign, and persuading teams that a smaller, honest figure is worth more than a flattering one.
See all articles by this author
