Juuuno
All research

Market structure

Paid on Spend: The Incentive Problem at the Heart of Ad-Tech

Most of the advertising supply chain is paid as a percentage of the money that flows through it. That creates a quiet but powerful incentive for more spend rather than better results. This article looks at how the incentive works, the evidence that it costs brands money, and what outcome-based alternatives look like.

Follow the fee, not the funnel

Every marketing team talks about the funnel: awareness at the top, purchase and loyalty at the bottom. Fewer talk about the fee structure that sits alongside it. Yet the way intermediaries are paid shapes what they optimise far more than any campaign brief.

In the programmatic supply chain, almost every layer is paid as a share of spend or a price per impression. The UK's ISBA/PwC study measured these layers directly [1]:

  • Agency fee: about 7% of advertiser spend
  • DSP fee: about 8%
  • Demand-side technology (serving, verification, data): about 10%
  • SSP fee: about 8%
  • Supply-side technology: about 1%

None of these fees is linked to whether the advertising worked. Each layer earns more when more money passes through it, and each is paid whether the resulting customers stay, buy or leave the next day.

Why the incentive matters

In most businesses, a supplier paid on volume is kept honest by the buyer's ability to measure what they received. Ad-tech makes that check hard, for three reasons.

The chain is long and opaque. The ISBA/PwC study could match only 12% of impressions end to end, and 15% of spend couldn't be attributed to anyone [1]. When the buyer can't see the whole chain, volume-based fees go largely unchallenged.

The metrics are the supplier's metrics. Impressions, clicks and installs are counted by the same systems that charge for them. They're easy to inflate and weakly linked to revenue.

The damage is delayed. A campaign that buys cheap, low-quality users looks good in week one and bad in month three. By then the budget has been spent and the fees paid.

The result is a market where it's rational for intermediaries to favour whatever grows spend: broader targeting, more inventory, more sites, more tools. It isn't necessarily bad faith. It's what the pricing model rewards.

The evidence that it costs brands

Waste shrinks when brands look. When the ANA began publishing data on made-for-advertising sites, low-quality inventory designed to soak up ad spend, advertisers responded quickly. The MFA share of spend fell from 15% in 2023 to 6.2% by late 2024 [2], and to a median of under 1% by 2025 [3]. That waste had been sitting in plain sight inside volume-priced chains.

Campaigns spread too thin. The ANA's 2023 study found the average campaign ran across 44,000 websites [4]. It later recommended consolidating to around 75–100 trusted sellers and five to seven SSPs [5]. A chain paid on volume has little reason to narrow its own reach.

Cutting spend doesn't always cut results. In 2017, Procter & Gamble cut about $200 million from its digital advertising, with reductions of 20–50% on several major platforms [6]. Marketing Week later reported that the reduction didn't reduce the effectiveness of P&G's advertising or affect sales [7]. When the world's largest advertiser can remove that much spend without losing results, some of the spend wasn't working.

The gap between leaders and the rest is large. In the ANA's Q1 2026 benchmark, top-performing advertisers turned 54% of programmatic spend into qualified impressions, while the rest managed 32%, and paid more than twice as much per qualified impression [8]. Better discipline pays, but most of the market doesn't have it.

"Upper-funnel" optimisation as a symptom

Volume-based pricing also explains a common frustration among performance marketers: campaigns that optimise for reach and clicks when the business needs paying customers.

Upper-funnel metrics are cheap to generate and easy to report. Lower-funnel metrics, such as purchases, retention and customer lifetime value (LTV), are harder to move and take longer to measure. An intermediary paid on spend has little reason to take on that harder work, because the fee is the same either way.

This is especially costly in app marketing, where the value of a user depends almost entirely on what happens after the install. A channel that delivers a million cheap installs with poor retention can look better on a volume report than one that delivers a hundred thousand users who stay and pay.

Automation doesn't fix incentives

The industry is automating fast. dentsu forecasts that 78.1% of ad spend will be algorithm-driven by 2027 [9]. But automation doesn't change who the algorithm works for.

An automated system optimises whatever goal it's given, inside whatever inventory it can access. If the goal is spend delivered or impressions served, automation simply delivers more of it, faster. If the system sits inside one platform, it optimises for that platform's inventory, not for the brand's best option across all of them.

Algorithms are an amplifier. Pointed at the wrong target, they amplify the wrong outcome.

What outcome-based pricing looks like

The alternative is to pay for results rather than activity. In marketing, outcome-based pricing usually takes one of three forms.

  1. Pay per verified action. Fees are linked to purchases, subscriptions or retained users rather than impressions or installs. This makes fraud and low-quality traffic much harder to profit from.
  2. Pay from return on ad spend (ROAS). The brand sets a ROAS goal, and the partner earns a share of the return achieved, paid from the campaign's profit. If the goal isn't met, the partner earns less or nothing.
  3. Hybrid models. A modest fixed fee covers the cost of running the service, and a performance component rewards results above a baseline.

These models aren't new in principle; affiliate marketing has long used cost-per-action. What's new is the ability to measure outcomes such as retention and LTV quickly and reliably enough to price on them, and to automate the optimisation that makes those outcomes achievable.

The objections, and the answers

"Outcome-based partners will only take easy wins." That's a risk, which is why goals should be set against a baseline and measured with incrementality testing, not just attribution.

"Outcomes take too long to measure." Early signals such as first purchase and early retention can serve as leading indicators, so pricing can be linked to them without waiting months for full lifetime value.

"Someone has to carry the risk." Yes. Under volume pricing, the brand carries all of it. Under outcome pricing, it's shared. That shift is the point.

Questions to ask any partner

  • How are you paid, and does your fee change if my customers don't stay?
  • What share of my spend reaches the publisher, and can you show me?
  • What are you optimising for: impressions, installs, or revenue and retention?
  • Would you accept part of your fee being linked to the ROAS you deliver?

The answers reveal more about a partner's incentives than any case study.

The takeaway

Ad-tech's biggest inefficiency isn't a technology problem. It's a pricing problem. When nearly everyone in the chain is paid on spend, the chain grows spend, and the evidence from P&G's cuts to the ANA's waste data suggests brands pay heavily for it. Fixing the leak means changing what the market pays for: outcomes, not volume.

References

[1] ISBA / PwC. Programmatic Supply Chain Transparency Study, executive summary. December 2020. https://www.isba.org.uk/system/files/media/documents/2020-12/executive-summary-programmatic-supply-chain-transparency-study.pdf

[2] ANA / TAG TrustNet. Programmatic Transparency Benchmark: 2024 study. December 2024. https://www.ana.net/content/show/id/pr-2024-12-programmatic

[3] ANA / TAG TrustNet. Programmatic Transparency Benchmark: Q2 2025. August 2025. https://www.ana.net/content/show/id/pr-2025-08-programmatictrans

[4] ANA. Programmatic Media Supply Chain Transparency Study: First Look. June 2023. https://www.ana.net/content/show/id/pr-2023-06-programmaticstudy

[5] ANA. Programmatic Media Supply Chain Transparency Study: complete report. December 2023. https://www.ana.net/content/show/id/83522

[6] Marketing Week. P&G's next frontier in digital marketing. March 2018. https://www.marketingweek.com/pgs-next-frontier-digital-marketing/

[7] Marketing Week. P&G on reach and ad spend. June 2019. https://www.marketingweek.com/pg-reach-ad-spend/

[8] ANA / TAG TrustNet. Programmatic Transparency Benchmark: Q1 2026. May 2026. https://www.ana.net/content/show/id/pr-2026-05-programmatic

[9] dentsu. Global ad spend forecast. June 2025. https://www.dentsu.com/news-releases/ad-spend-forecast-to-grow-by-four-point-nine-percent-in-2025-despite-a-reduced-economic-outlook

Back to Juuuno