Agentic Advertising: Someone Has To Carry The Proof Gap
A DMEXCO column by Evgeny Popov on agentic advertising: AI agents can stay within budget and still spend too much, because the money moves before the proof arrives. Who carries that gap – and what would it take to price it?
Agentic Advertising: Staying In Budget Is Not Enough
In agentic advertising, an AI agent can stay inside its budget and still spend too much.
It can buy on signals from known sources. It can act within the authority it was given. It can log every decision. And the evidence can still arrive after the money is committed.
In three earlier pieces I argued that agents need richer signals than segments, feedback fast enough to reprice the next impression and standards that establish authority rather than parse data. Grant all three. There is still a hole in the plan, and it has a dollar figure.
What Is The Proof Gap In Agentic Advertising?
The proof gap is spend committed while the evidence needed to assess a change is still pending.
Suppose a brand hands an agent $200,000 and the agent moves $1,000 an hour into a new mix of connected TV (CTV) and retail media. The first sales readout fit to judge that change takes 48 hours to mature. By the time it lands, $48,000 has been committed to a decision nobody has assessed – and that first readout will not settle all $48,000 either. It will produce an estimate, with error bars, about a change that is already two days old.
That gap is not a waste. The change may be right. But it is exposure, and it is a line most media plans leave implicit. The campaign budget caps total spend. Pacing sets the rate. Neither, by itself, limits how much can ride on a single untested change while the test is open.
Faster measurement narrows the gap but never closes it. A model updates its estimate in milliseconds; a business outcome takes weeks to become real. And in an agentic system the gap compounds. One agent changes bids, audiences and channel mix at once, so the same weak guess drives spend across several budgets. Each move looks small. The total at risk does not.
Who Holds The Proof Gap?
This industry has fought over the space between commitment and settlement before. In the early 1990s the fight was sequential liability: whether an agency owed the publisher before the client had paid the agency. That was a fight about credit risk – who eats the loss if the money never arrives.
The agentic version is about proof risk – who eats the loss if the outcome never arrives. Same question, one layer deeper – and the terms, when someone writes them, will have to distinguish evidence that is missing from a shortfall that has been measured.
Markets that trade ahead of proof lean on two tools. A limit on what you commit while you learn. And a contract that assigns a defined loss to someone else when a defined thing goes wrong. They are different instruments, and the mistake this industry is about to make is treating them as one.
What Is A Risk Budget For AI Agents?
The first tool is a second budget line. Alongside the campaign budget, the brand sets a risk budget: the most an agent may commit under an unvalidated change before it must show the agreed evidence or obtain fresh approval. In the example, the agent might be granted $12,000 to test the new mix – if $12,000 can fund a test with enough data to guide the next move – and nothing more until the readout clears a threshold both sides set in advance.
Three rules keep it honest.
- The evidence standard is agreed before the first bid: what counts as progress, which method measures it and how much of it expands the agent’s scope. A pixel records an event. A sound causal test estimates how much the ads changed outcomes, with stated uncertainty. The agent’s own confidence score is not independent validation.
- The clock matches the goal. A car, a mortgage and a brand campaign do not prove themselves in 48 hours, and a short-term sales gate should not be the default for work designed to pay back over months.
- And when evidence is late, the default is pause, not expansion. A lack of proof never grants fresh funds.
One more rule, and it is the one an optimizer will test first: the risk budget has to sit outside the agent’s objective. A ceiling the agent is rewarded for reaching stops being a ceiling. The agent cannot raise the cap, and it cannot reset the count by moving spend into a new channel or handing it to a second agent. It is a constraint on the delegate, not a target for it.
The Transfer: An Underwriter For The Miss
A risk budget limits what the buyer commits while it learns. A guarantee does something different: it assigns a defined financial loss to a counterparty when an agreed condition is met. For that to hold, the contract has to say what triggers payment, how much is owed, what is excluded and who funds the promise. The test needs a limit. The guarantee needs its own.
The first buyer of this cover may be the seller. A brand can halt new spend, though it cannot recover money already committed. A publisher still needs payment for the media it has delivered. An intermediary may have bought impressions that fail to earn back their cost. Those risks need separate terms. The seller may buy the cover to support the promise it makes to the brand. The buyer’s test allowance still stays in force.
Advertising has one working precedent for the form. The make-good is a contractual remedy for a defined shortfall in delivery: under-deliver the impressions, cure with inventory. The edges of that form are being probed now. Munich Re has backed AI vendors’ performance guarantees with a reinsurer’s balance sheet since 2018. TikTok offers ad credits to eligible campaigns on GMV Max, its ad product optimized for gross merchandise value (GMV), when ROI falls below 90% of the daily target. Both have a defined trigger and a defined payout. One is third-party capital paid in cash. The other is the seller’s own count, paid in credits for future ads, behind a long list of exclusions – including the day a campaign is paused. The stop rule and the cover have to work together.
Now hold everything else constant. Two sellers pitch the same outcome deal at the same cost per acquisition (CPA), counted by the same agreed method, billed on the same terms. One offer stops there. The other adds a funded, capped cash remedy for a defined shortfall, at a stated fee. The test allowance does not move; the buyer needs the same evidence either way. What moves is the buyer’s retained loss if the miss happens. Coverage does not validate the strategy. It changes who eats the miss – and the fee for that is the price of the specified risk being transferred.
Its value depends entirely on the terms, the price and the strength of the party behind it. That is not a caveat; it is the point. A guarantee without reserves behind it is a slogan.
A New Basis For Fees
This also creates a different basis for fees. A vendor paid a share of spend earns less every time its agent pauses, which asks it to prize restraint while paying it for volume. The intermediary of the agentic era can be paid the other way around: a spread for standing between two parties when the thing they agreed on does not happen. Finding counterparties, negotiating price and running the optimization are information work, and software is driving the cost of information work toward zero. Software can price the risk. A funded counterparty still has to bear the loss.
I expect the first funded promises to cover counting errors. Covering poor business results is a further step.
My prediction: by the end of 2027, at least one media deal between independent firms will settle from escrow. Its terms will name the source used to count results and provide a cash remedy for a counting error. The test is an executed deal with terms that can be verified. If none can be verified by that date, the call has missed.
Closing The Loop
In my last piece I argued that AI can interpret data but cannot vouch for it. Neither can capital, and the distinction matters. A test estimates what the spend did. A contract says who answers if it didn’t. A balance sheet says whether that answer is worth anything.
The standard of proof belongs in the test. The promise to pay belongs in the contract. The cash behind the promise belongs on someone’s balance sheet.
The right to spend more should be earned. Any promise to cover the miss should be priced, bounded and backed.
