On paper, paying an agency for results is the fairest arrangement there is. In practice, many results-based contracts end in an argument about whose number is right. The cause is rarely bad faith. Usually the deal was signed before anyone agreed how a result would be counted, which season it would be compared with, or which customers should qualify. Here are the three failure points to close before launch.

Why performance-based marketing deals break down

The ANA reports that use of performance incentives fell from 61% of advertisers in 2013 to 15% in 2025. That does not prove paying for results is a bad idea. It shows that agreeing on results is hard: in a June 2026 report cited by ANA Magazine, 81% of agencies saw the appeal of outcome-based models and 69% named attribution as the blocker.

Failure 1: A number neither side controls

Each ad platform reports the revenue it believes it caused, inside its own attribution window, so the same order can be claimed twice. Northbeam, an attribution vendor, states plainly that its numbers may not line up with Meta Ads Manager by design. The gap does not always flatter the platform, either. Across 640 incrementality experiments, Haus found Meta’s 7-day click reporting under-counted measured incrementality by 15% on average for brands selling only on their own site. A fee tied to any single platform’s report is tied to a number neither side controls.

Failure 2: No baseline

Revenue rises in November whatever the agency does, and a strong product launch lifts every channel. If the contract pays on total revenue without a baseline, the agency is paid for the calendar. Compare against the same months of the prior year, and write that comparison into the agreement.

Failure 3: Paying for the wrong customer

A revenue share that counts returning customers rewards retargeting people who would have bought anyway. Branded search is the classic example: it converts well and is cheap to claim. Either exclude these sources or pay a separate, lower rate for them, so the bonus follows new customers rather than easy credit.

Fix the measurement before you sign

Forrester has argued that performance-based agency compensation, as usually designed, does not produce better work from agencies. Treat that as a fair warning rather than a verdict: solve the measurement first, then decide whether to pay on it. Tracking has to be trustworthy before any fee depends on it, because a broken Pixel can make a good month look bad, or the reverse. That is the work behind our tracking and attribution service, and you should keep the ad accounts, Pixel and data in your own name so the measurement does not leave with the vendor.

Conclusion: Make Performance-Based Marketing Deals Survive a Bad Quarter

Performance-based marketing deals fail on measurement, baseline and customer definition, not on the idea of paying for results. Close those three gaps in writing before the first campaign goes live. To see which of your numbers can support a results-based fee, book a free 30 minute growth audit, or read the full guide to how a performance-based marketing agency prices pay-for-performance for ecommerce.