The Cost of Renting Ad Tech: Annual Report Methodology

The Cost of Renting Ad Tech is an annual report that combines public supply-chain studies, company filings, cloud and labor price series and anonymized platform data into one set of metrics: fees per dollar, working media, effective take rates, fee variance and the cost of owning the stack at three budgets. This page is the method; first edition March 2027.

The report exists because the public evidence is scattered across studies with different metrics, countries and years. Putting them on one page with their definitions, and adding what filings and price lists say, gives a number that can be cited with its caveats attached.

Platform data is the only input no one else has, and the one readers will trust least, so the rules are strict and public.

1. Customers opt in, in writing, per edition, and can withdraw any time before publication.
2. Aggregates are computed inside each customer's environment and exported as counts and sums; no bid-level or log-level data leaves the customer.
3. A published cell needs at least five contributors, and no contributor may account for more than 40% of a cell's volume.
4. Fee percentages are rounded to the nearest 0.5 point and CPMs to $0.10; volumes are shown as ranges.
5. Data is at least 90 days old at publication.
6. Cells are defined by region (US, UK, EU) and channel (display, video, CTV, retail media); any cell below the thresholds is suppressed, and suppression is shown as "n/d", not blank.
7. The aggregation code is published with the edition and reviewed by an external engineer named in the report.
8. No customer is named, and no figure is presented in a way that allows a customer to be inferred.
9. Platform-derived figures are always shown beside a public figure for the same metric and labelled "platform data, n contributors".

Every metric is reported with its source, date, country and sample, and metrics from different definitions are never averaged together.

1. Headline: the fee share per dollar this year, last year, and the five-year view.
2. Method and sources (this page, condensed).
3. The dollar's journey: a waterfall from advertiser spend to publisher receipt, US and UK, with each study's figure.
4. Working media over time: the ANA series, with the metric changes marked.
5. Take rates from filings: DSP, SSP and retail media platforms, with the calculation shown.
6. Fee variance: what the same inventory costs through different paths.
7. The unattributable gap and what closed it.
8. Retail media: gross billings to net revenue.
9. What owning costs: TCO at three budgets; infrastructure per billion requests; staff.
10. Platform data: the opt-in aggregates, with contributor counts.
11. Limitations.
12. Appendix: source register, data dictionary, aggregation code, "cite as".

All charts: SVG with embed code, CC BY 4.0 attribution, n and source printed on the chart.

• Annual edition each March, after the Q4 ANA release (the Q4 2025 findings came on 23 February 2026, ANA library) and the 10-K season (The Trade Desk filed in February 2026). First edition: 16 March 2027.
• Mid-year data note each September, after the ANA Q2 release (6 August in 2026), updating charts 3 and 10 only.
• Editions live at /research/cost-of-renting-ad-tech/{year}; the canonical URL shows the latest with links to prior years. Each edition has a DOI and a data file; prior editions are never edited, only annotated.
• Methodology versioning as for the TCO methodology: major for metric definition changes, minor for source changes, patch for text. Metric definitions changed in a major version are reported both ways for one edition.

• The public studies measure different things: the ANA 2023 study measured spend reaching the consumer, the quarterly benchmark measures spend reaching publishers, and TrueAdSpend adds viewability, validity and MFA filters. The report keeps them on separate lines and says so on every chart.
• ISBA's figures are UK, 2020 and 2022, from 15 advertisers; the ANA's are US. Neither is a probability sample of the market, and participants are large advertisers that chose to be studied.
• The Trade Desk ratio includes the platform fee in gross spend and bundles data and services, so it overstates the pure media fee; it is one company.
• 10-K segment figures mix lines (Criteo's media spend covers retail and performance media; Walmart records advertising in net sales or as a cost-of-sales reduction).
• Adalytics figures are from specific investigations, not a market sample, and are cited through trade press where the original report is not public.
• eMarketer figures are forecasts and are labelled as such.
• Cloud prices are US-region list prices; committed and reserved pricing is lower, EU regions higher. The wage series is US and for software developers.
• Platform data comes from the sponsor's customers, who are not representative of the market and who chose an in-house model; it is shown beside public figures and never as a headline.
• Rebates, kickbacks and agency principal trading are not visible in any of the sources and are outside the report's scope.
• Currency: UK and EU figures are reported in their own currencies and not converted.

This report measures fees. Its companion, State of AI in Programmatic 2027, measures practice: which bidding, planning and agent tools practitioners use and what they expect. Both are released with open data under CC BY 4.0 on the same cadence so the two can be read together.

How much of every advertising dollar reaches the publisher? About half, depending on the study and the year. ISBA and PwC found 51% in the UK in 2020 (Marketing Dive, 7 May 2020); the ANA benchmark reported 47.1% reaching publishers in Q3 2025 (ANA, 5 November 2025). Stricter measures that remove unviewable, invalid and MFA impressions are lower: 36 cents in the ANA's 2023 study and a 45.1% TrueAdSpend index in Q2 2026 (ANA, 17 June 2024; Fiducia on ANA, 6 August 2026).

What is the ad tech tax? The share of spend retained by intermediaries between advertiser and publisher: DSP, SSP, exchange, data, measurement and verification fees, plus whatever cannot be attributed. The ANA put transaction costs at 29% of each dollar in 2023 and 29.8% in 2024 (ANA, 17 June 2024; ExchangeWire, 16 September 2026). The ad tech tax explained page covers the layers.

Are ad tech fees going down? The evidence says working media is rising and the unattributable gap has shrunk, while fee percentages are flat. ISBA's unknown delta fell from 15% to 3% between 2020 and 2022 (ExchangeWire, 16 September 2026); the ANA's transaction-cost figure moved from 29% to 29.8% (ANA, 17 June 2024; same ExchangeWire piece); The Trade Desk's revenue over gross spend rose from 20.3% to 21.6% (The Trade Desk 10-K, FY2025, our calculation). Improvements have come from media quality, not from cheaper intermediaries.

How is this report different from the ANA benchmark? The ANA benchmark is log-level data from member advertisers, quarterly, US, behind membership for the full reports. This report is annual, public, combines the ANA and ISBA series with filings, price lists and platform data, adds the cost of owning the stack, and releases its data. It cites the ANA rather than competing with it; the working media benchmark page tracks the ANA series quarterly.

Where does the platform data come from and can it be trusted? From customers of the sponsor who opt in per edition, aggregated inside their own environments under the nine rules above, with a minimum of five contributors per published cell and an external review of the aggregation code. It is shown beside public figures and never used alone for a headline. Readers who distrust it can ignore chart 11 and lose nothing else in the report.

Use the in-house programmatic guide to compare the related operating models, definitions and evidence before making this decision.

1. ANA, "ANA, TAG TrustNet Launch Programmatic Transparency Benchmark" (17 June 2024). https://www.ana.net/content/show/id/pr-2024-06-tag
2. ANA, Programmatic Transparency Benchmark library (reports from 5 December 2023 to 26 August 2026). https://www.ana.net/micpackage/show/id/97114
3. ANA, Q3 2025 Programmatic Transparency Benchmark release (5 November 2025). https://www.ana.net/content/show/id/pr-2025-11-transparency
4. Fiducia, "ANA Programmatic Transparency Benchmark Q2 2026: working media efficiency reaches a record 45.1%" (6 August 2026). https://www.fiducia.eco/post/ana-programmatic-transparency-benchmark-q2-2026-working-media-efficiency-reaches-a-record-45
5. Marketing Dive, "15% of programmatic ad spend disappears in the supply chain, study says" (7 May 2020). https://www.marketingdive.com/news/15-of-programmatic-ad-spend-disappears-in-the-supply-chain-study-says/577525/
6. ExchangeWire, "The Ad Tech Toll Road" by Shirley Marschall (16 September 2026). https://www.exchangewire.com/blog/2026/09/16/the-ad-tech-toll-road/
7. Adalytics, research blog (accessed 17 October 2026), including "On pre-bid bot detection and filtration". https://adalytics.io/blog
8. The Trade Desk, Form 10-K FY2025. https://www.sec.gov/Archives/edgar/data/1671933/000167193326000014/ttd-20251231.htm
9. Criteo S.A., Form 10-K FY2025. https://www.sec.gov/Archives/edgar/data/1576427/000157642726000014/crto-20251231.htm
10. Cardlytics, Form 10-K FY2025. https://www.sec.gov/Archives/edgar/data/1666071/000166607126000010/cdlx-20251231.htm
11. Marketplace Pulse, "Amazon Advertising Services Sales" (accessed 17 October 2026). https://www.marketplacepulse.com/stats/amazon-advertising-services-sales
12. Walmart, Q4 and full-year fiscal 2026 earnings release (19 February 2026). https://stock.walmart.com/sec-filings/all-sec-filings/content/0000104169-26-000032/earningsreleasefy26q4.htm
13. Instacart, First Quarter 2026 Financial Results (6 May 2026). https://seekingalpha.com/pr/20502657
14. Vantage, EC2 c7g.xlarge (updated 17 October 2026). https://instances.vantage.sh/aws/ec2/c7g.xlarge
15. CloudZero, "AWS egress costs" (updated 1 May 2026). https://www.cloudzero.com/blog/aws-egress-costs
16. Google Cloud, VPC pricing announcement (effective 1 February 2024). https://cloud.google.com/vpc/pricing-announce
17. US Bureau of Labor Statistics, Software Developers (May 2025; modified 27 August 2026). https://www.bls.gov/ooh/computer-and-information-technology/software-developers.htm
18. Hetzner, AX102 dedicated server (accessed 17 October 2026). https://www.hetzner.com/dedicated-rootserver/ax102/

• 18 October 2026: aligned the title and heading and applied recorded source-URL corrections where needed. Factual research cutoff remains 17 October 2026.

• 1.0.0, 17 October 2026. Methodology published: 10 sources in the register, 11 metrics, 12 charts, 9 disclosure rules. First edition scheduled 16 March 2027.