The Price of an Outcome
The industry has agreed to sell outcomes. Nobody agreed on how to price one. Here is the arithmetic.
Every vendor in commerce now sells outcomes. Not impressions, not seats, not hours — outcomes. Revenue lift. Acquired customers. Incremental margin. It is the right direction for the industry, and it rests on an assumption almost no buyer can actually satisfy: that you know what an outcome costs you.
You can't buy an outcome you can't price. This piece is about learning to price one.
Where the dollar actually goes
Start with the plumbing, because the plumbing is where the money disappears.
Take a commerce marketing dollar and follow it from your budget line to the moment it does productive work. It passes through agency fees. Platform fees. Ad-tech intermediaries, each taking a metered slice. Data and measurement tooling. And now — the newest tranche, growing faster than any other — the AI subscriptions layered on top of all of it. What remains at the end of that chain is your working money: the portion of the dollar that actually reaches the activity you thought you were buying.
The figures vary by operation, which is exactly the point — the only version of this waterfall that matters is yours, reconciled to your accounts. But the shape is consistent, and it is consistently worse than the buyer assumes. (In our prior firm's programmatic engagements — real numbers, from real enterprise waterfalls — working media routinely started at a fraction of the dollar, and disciplined restructuring lifted it fourfold.) Any illustrative figure you see in our materials is marked as illustrative; the waterfall you should trust is the one built from your own cost structure.
The waterfall is the first half of pricing an outcome: it tells you what a dollar of input is actually worth by the time it goes to work. Most operations have never drawn it. Every Pareto engagement starts by drawing it.
What a customer actually costs
The second half is the output side: what did the outcome cost, measured honestly?
Ask three platforms what a customer costs and you will get three confident, mutually incompatible answers — each one computed on the platform's own attribution logic, each one grading its own homework. Platform-reported CAC is not a measurement. It is a sales document.
Real CAC is computed from financial ground truth: total spend by channel — all of it, including the fees and tooling the waterfall surfaced — over customers actually acquired, reconciled to the systems where money and orders actually live. Run that calculation and two things reliably happen. The blended number is higher than any platform reported. And the ranking of channels changes — sometimes dramatically — because channels differ enormously in how much invisible cost rides on top of their visible spend, and in how much of their claimed credit survives contact with incrementality.
Real CAC by channel is the single most clarifying artifact a commerce operation can build. It converts "which channels are working?" from a debate between dashboards into a line of arithmetic. And it is the number an outcome has to beat: an acquired customer is only an outcome if it was acquired below the margin it contributes.
What outcome-based pricing actually requires
Now the pricing conversation can be had honestly. Outcome-based pricing — paying vendors, agencies, or AI systems on results — requires three things, and the industry conversation routinely skips all three.
First, a defined outcome with its success rule written before the spend, not discovered afterward in whatever metric happens to look good. Second, a causal standard: the outcome must be incremental — something that happened because of the spend, not merely near it. Attribution proximity is not causation, and paying on it means paying for outcomes that were coming anyway. Third — and this is the one the industry structurally avoids — a referee.
Here is the problem with the referee. In most outcome deals, the party reporting the outcome is the party being paid on it. The platform measures its own lift. The agency reports its own performance. The AI vendor's dashboard grades the AI vendor's work. Every one of those parties may be honest, and it does not matter: a scorekeeping system where the players keep score is not a scorekeeping system.
The referee has to be independent — measuring from financial ground truth, reconciled to the bank, with no economic stake in the score. This is why Pareto never bills as a percentage of spend: a firm paid more when you spend more cannot be trusted to tell you to spend less. We keep score; we are not paid on the score.
The sequence
So the method, in order: draw the waterfall, so you know what your input dollar is worth. Compute real CAC by channel, so you know what your outcomes actually cost. Then — and only then — entertain outcome-based deals, with success rules written up front, incrementality as the standard of proof, and a referee with no skin in the score.
Outcomes are the right thing to buy. Just price them first.
