← All articles

How Much of a Programmatic Ad Dollar Reaches a Real Ad? 18 Sources, 2020 to 2026

This analysis draws on data from 18 credible sources, including government and court findings, industry reports and company data.

In 2023, only 36 cents of each dollar an advertiser put into a demand-side platform reached a consumer.2 By the second quarter of 2026, the same benchmark measured 45.1%.8

That is real progress, and it still means more than half of programmatic spend does not become a viewable, measurable, fraud-free ad on a quality site. The data from 2020 to 2026 also shows where the gain came from. Fees barely moved. Waste fell.

We'll explore:

Where the money went in 2023

The Association of National Advertisers (ANA) ran the largest audit of open web programmatic buying to date. It covered 21 member companies, $123 million in spend and 35.5 billion impressions, bought between September 2022 and January 2023.1

The full report split each DSP dollar three ways. 36% reached the consumer. 29% went to transaction costs, including DSP and SSP fees. The other 35% bought impressions that were non-viewable, invalid, non-measurable or on made-for-advertising (MFA) sites.2

The ANA sized the open web programmatic market at $88 billion.1 It put the waste at as much as $20 billion in its first look, 23% of spend.1 The complete report raised the efficiency opportunity to $22 billion, one dollar in four.2

Two numbers from the first look explain most of that loss. Campaigns ran on an average of 44,000 websites.1 And MFA sites, built to harvest ad impressions rather than readers, took 21% of impressions and 15% of spend.1 The ANA's recommendation was 75 to 100 trusted sellers and five to seven SSPs.2

Figure 1: Each Programmatic Dollar, 2023 vs Q2 2026 (%)

Source: 2, 8.

Fees have barely moved since 2019

Every independent measure of the intermediary cut lands in the same range.

The first ISBA and PwC study traced UK programmatic spend from advertiser to publisher. Publishers received an average of 51%.9 A further 15% could not be attributed to anyone, about a third of supply chain costs.9 Of more than 1,000 supply chains, only 290 could be matched end to end.9

The UK Competition and Markets Authority reached a similar answer the same summer. It found intermediaries capture at least 35% of the value of advertising bought from newspapers and other content providers.11 It estimated that publishers receive around 65% of what advertisers pay, and said the true ad tech take is likely higher than 35%.11

Australia's competition regulator measured fees directly. The four key ad tech services took 27% of programmatic spend in 2020, after 28% in 2019.12 The ACCC said the real total is likely higher, because buyers and sellers often pay for services beyond those four.12

The largest intermediary's own figure fits. Google said publishers using Ad Manager kept over 69% of what advertisers paid through Google Ads or Display & Video 360 in 2019.18 Its standard DV360 rate is 15%.18 In April 2025 a US federal court found that Google's durable ability to charge a 20% take rate on its AdX exchange showed monopoly power, and that Google held more than 90% of the publisher ad server market.13 The ACCC had found that over 90% of ad impressions traded through the Australian ad tech supply chain in 2020 passed through at least one Google service.12

Six years later the ANA's own fee line had moved two points. Transaction costs were 29% of spend in 20232 and 27.2% in the second quarter of 2026.8

Figure 2: Share of Programmatic Spend Taken by Fees and Intermediaries, by Study (%)

Source: 11, 2, 12, 8, 13.

What did change is how much of the chain anyone can see. ISBA's second study cut the unattributable share from 15% to 3%.10 The impression match rate rose from 12% to 58%, and the share of spend reaching publishers rose by 8%.10 The work took nine months instead of 18.10

The gain came from cutting waste

The ANA turned its one-off study into a quarterly benchmark, built with TAG TrustNet and Fiducia.8 Its Q1 2025 results showed participants directing 41% of budgets to effective impressions, against 36% in 2023.3

The second quarter of 2025 showed how. Median MFA spend fell from 2.3% to 0.8%, against 15% two years earlier.4 Active domains fell from 53,799 to 28,958, nearly half.4 The median SSP count fell from 19 to 17.4 Private marketplace deals rose from 64.5% of spend to 87.8%.4 The ANA still counted $26.8 billion in global media value lost each year to inefficiency.4

The private deals cost more per impression. Average CPMs were $7.15 in private marketplaces against $4.41 in the open marketplace.4 Buyers paid more per thousand and lost less of it.

By the third quarter, the share of spend delivering fraud-free, measurable, viewable, MFA-free impressions was 39%, two points up on the second.5 Non-measurable impressions fell from 28% to 9.2%.5 MFA was down to 0.39% of spend, and 99.1% of spend ran in low-risk environments.5 The ANA valued the working media recovered at $13.6 billion.5

The index reached 43.3% in the first quarter of 20267 and a record 45.1% in the second.8 Non-viewable spend fell from 13.3% to 10.1% that quarter, and invalid traffic was 0.1%.8

Figure 3: Share of Programmatic Spend Reaching a Quality Impression, ANA (%)

Source: 2, 3, 5, 7, 8.

Not every quality measure kept improving. MFA exposure rose to 1.1% in the first quarter of 2026, from 0.4% to 0.6% through 2025.7 The benchmark also changed shape as it grew: participating marketers rose from 54 to 86 that quarter.7

Fraud is low where it is policed

Invalid traffic was 0.1% of the ANA's tracked spend in mid-2026.8 Industry fraud data outside the benchmark shows that figure depends on where the money runs.

In the US, the Trustworthy Accountability Group has measured invalid traffic below 1% in its certified channels for four straight years.15 That study counted every impression from participating agency holding companies in the first half of 2024, and certified channel volume has grown from 6.5 billion impressions in 2017 to 745 billion.15 TAG credits industry anti-fraud programs with saving US advertisers $10.8 billion in 2023.15

Europe shows the other side. Without current standards, TAG's 2025 study estimated invalid traffic in display and video at about 10.36%, roughly €5 billion in 2023.16 In the 76% of spend that ran through protected channels, losses were 1%.16 The other 24% ran through channels without those standards.16

Viewability is the larger leak. Integral Ad Science measured global display viewability at 67.9%, against 79.7% for video.17 Mobile web display was 45.1% of impressions but 54.9% of brand suitability violations and 71.9% of MFA impressions.17

The gap between buyers is now the story

The market average hides two very different groups of buyers.

In the fourth quarter of 2025, advertisers with disciplined quality controls turned 56.7% of programmatic spend into qualified impressions. Lower performers turned 37.5%.6 The share of advertisers above 50% fell from half to 43.8%.6 Case studies showed cost per conversion falling by nearly 40% when buying was optimized to quality-adjusted metrics instead of CPM.6

In the first quarter of 2026 the split was 54.0% against 32.1%. At 21.9 points it was the widest gap the benchmark had recorded.7 Lower performers lost 38.4% of spend to media quality problems, twice the higher group's 19.0%.7

Fees do not explain the gap. Transaction costs differed by 2.4 points between the two groups. Media productivity losses differed by 19.4.7

The price per useful impression shows it best. Higher performers paid $7.46 per thousand qualified impressions and lower performers paid $19.04.7 A $1.95 difference in the CPM they paid became an $11.58 difference once waste was counted.7 Overall CPMs fell from $5.55 to $4.42 that quarter.7 A cheaper CPM is not a cheaper ad.

The second quarter of 2026 repeated the pattern at 52.3% against 31.1%.8 Lower performers lost 22.7% of spend to non-measurable inventory, more than double the 10.4% lost by higher performers.8 Higher performers bought across about 19,000 fewer domains and apps.8

Figure 4: Spend Converted to Qualified Impressions, Higher vs Lower Performers (%)

Source: 6, 7, 8.

Why the percentages matter at this scale

Programmatic is now most of digital display and video buying. US programmatic ad revenue reached $162.4 billion in 2025, up 20.5%, out of $294.6 billion in total internet advertising.14 Non-programmatic placements outside search fell 13.9% to $18.0 billion.14 Programmatic was $99.0 billion in 2021.14

More of it is moving to connected TV. CTV's share of the ANA benchmark's programmatic spend rose from 30.4% to 44.2% in the second quarter of 2025.4 It was 45.6% in the third quarter5 and 41.6% in the second quarter of 2026.8 In the fourth quarter of 2025, every CTV transaction in the benchmark ran through a private marketplace.6

At that scale, each point of working media is a large sum of money. The ANA's own estimates of lost value ran from $20 billion in 20231 to $26.8 billion in mid-2025.4

What the data does and does not show

Taken together, the 18 sources support three findings.

The intermediary cut is stable and large. Regulators in two countries, an audit of UK supply chains, the ANA and Google's own disclosure put it between roughly 27% and 35% of spend from 2019 to 2026.11,12,2,8,18 Transparency improved far more than price.10

Working media rose because buyers cut waste. Fewer domains, fewer SSPs, private deals and less MFA moved the ANA figure from 36% to 45.1%.2,4,8

The average now matters less than the buyer. Between the better and worse halves of the benchmark, the difference in working media is about 20 points, and almost all of it is media quality rather than fees.7,8

The data has limits. The ANA benchmark is a self-selected group of large advertisers, and its participant base grew from 54 to 86 marketers in one quarter.7 Its headline measure has changed name and definition across releases, so quarter-to-quarter moves of a point or two should be read loosely. The fee estimates come from different countries and years and do not measure the same layers of the chain. No source here measures small advertisers buying through self-serve platforms, so whether they keep more or less than 45 cents is not known.

Sources

  1. ANA. Programmatic Media Supply Chain Transparency Study: First Look. June 19, 2023.

  2. ANA. Programmatic Media Supply Chain Transparency Study: Complete Report. December 5, 2023.

  3. ANA. Programmatic Transparency Benchmark: Q1 2025 Findings. May 21, 2025.

  4. ANA. Q2 2025 Programmatic Transparency Benchmark. August 14, 2025.

  5. ANA. Q3 2025 Programmatic Transparency Benchmark. November 5, 2025.

  6. ANA. Q4 2025 Programmatic Transparency Benchmark. February 25, 2026.

  7. ANA. Q1 2026 Programmatic Transparency Benchmark. May 27, 2026.

  8. ANA. Q2 2026 Programmatic Transparency Benchmark. August 6, 2026.

  9. ISBA and PwC. Programmatic Supply Chain Transparency Study. May 2020.

  10. ISBA. Second Programmatic Supply Chain Transparency Study. January 18, 2023.

  11. UK Competition and Markets Authority. Online platforms and digital advertising market study: final report. July 1, 2020.

  12. ACCC. Digital advertising services inquiry: final report. September 28, 2021.

  13. United States v. Google LLC (E.D. Va.). memorandum opinion, key excerpts. April 17, 2025.

  14. IAB and PwC. Internet Advertising Revenue Report: Full Year 2025. April 16, 2026.

  15. TAG. US Fraud Benchmark Report. October 29, 2024.

  16. TAG. 2025 European Ad Fraud Savings Report. June 17, 2025.

  17. Integral Ad Science. 21st Media Quality Report. July 9, 2026.

  18. Google. How our display buying platforms share revenue with publishers. June 2020.