Market sizing
Sizing the Leak: How Much Global Digital Ad Spend Never Reaches a Real User?
Combining global market data with audited supply-chain costs and fraud estimates, we estimate that around $207 billion of the roughly $790 billion spent on pure-play digital advertising in 2025 was lost before it reached a real user. That's about 26 cents of every digital dollar. This article sets out the model, its assumptions and where it could be wrong.
Why size it at all?
Individual studies tell us that a large share of the programmatic dollar is lost in the supply chain. What they don't tell us is how big the problem is across all digital advertising. That number matters. It shows how much value is available to anyone who can fix the problem, and it gives brands a way to think about their own exposure.
Sizing it honestly means resisting a tempting shortcut. It would be easy to take the headline finding that only around 45 cents of an intermediated dollar reaches a real user and apply it to the whole market. That would be wrong, and it would overstate the problem.
Step 1: how big is digital advertising?
WPP Media's This Year Next Year midyear 2025 forecast put global advertising revenue at $1.08 trillion for 2025, with pure-play digital channels accounting for 73.2% [1]. That puts pure-play digital at roughly $790 billion.
Other sources use different definitions. WPP's December 2025 update revised 2025 revenue to $1.14 trillion [2], and its midyear 2026 update forecast about $1.3 trillion for 2026 [3]. dentsu, measuring differently, put 2025 global ad spend at $992 billion, with digital at $678.7 billion, or 68.4% [4]. For consistency, this model uses WPP's midyear 2025 total and share together rather than mixing sources.
Step 2: split the market
Not all digital spend passes through the same kind of chain.
Transparent engines. A large share goes directly to self-serve platforms that measure and report their own results: Google, Meta, Apple, Amazon, TikTok and device makers (OEMs). A brand buying on these platforms doesn't pay a separate DSP, SSP and exchange. It may pay an agency fee on top.
The intermediated layer. The rest flows through open-web programmatic, ad networks, affiliates and resold direct deals. This is where the layered fees, untraceable spend and fraud described in the ISBA/PwC study [5] and AppsFlyer's data [6] concentrate.
There's no single published figure for how global digital spend divides between these two groups. For our base case we use a 60/40 split (60% transparent, 40% intermediated), based on operating experience across the brands we work with. Because this is the input with the most uncertainty, we test a range from 30% to 50% intermediated below.
Step 3: apply a leak rate to each part
Intermediated layer: 55%. This comes from the 45-cent model: the fees and unknown delta measured by ISBA/PwC [5] (49% combined), plus fraud at about 12% of the working media that remains, in line with AppsFlyer's global estimate [7].
Transparent engines: 7%. Here we count only the typical agency fee measured in the ISBA/PwC study [5]. We deliberately don't count fraud inside the large platforms, even though some vendors report significant levels there. The mFilterIt ad fraud report index, for example, notes fraud of 9–18% in walled gardens [8]. Leaving it out makes the estimate conservative.
Step 4: the result
| Component | Spend | Leak rate | Lost per year |
|---|---|---|---|
| Transparent engines (60%) | $474B | 7% | $33B |
| Intermediated layer (40%) | $316B | 55% | $174B |
| Total | $790B | ~26% | ~$207B |
On these assumptions, about $207 billion a year, or roughly 26 cents of every digital dollar, is lost before it reaches a real user.
Sensitivity: what if the split is different?
| Intermediated share | Intermediated leak | Transparent leak | Total |
|---|---|---|---|
| 30% | $130B | $39B | $169B |
| 40% (base case) | $174B | $33B | $207B |
| 50% | $217B | $28B | $245B |
Even at the low end, the loss is well over $150 billion a year.
A sanity check against fraud alone
A useful test of any estimate like this is whether it's consistent with other independent numbers. Juniper Research estimated that fraud alone cost advertisers about $84 billion in 2023 [9]. Our $207 billion covers fees, untraceable spend and fraud together, so it should be meaningfully larger than the fraud figure alone. It is.
The app-marketing slice
App marketing is a useful sub-case because the data is unusually detailed. AppsFlyer reported that global spend on app user acquisition rose 13% to $78 billion in 2025, with non-gaming apps accounting for $53 billion and gaming for $25 billion. Remarketing spend grew 37% to $31.3 billion [10]. Together that's roughly $109 billion of app marketing, all of it inside the $790 billion total.
If 30–50% of user-acquisition spend flows outside the large self-reporting networks, and 55% of that leaks, then $13–21 billion of app user-acquisition budget is lost every year.
India: the home market
In India, the dentsu-e4m Digital Advertising Report 2026 put digital ad spend at INR 71,621 crore in 2025, 59% of a INR 1,21,339 crore market [11][12]. Programmatic accounted for INR 30,081 crore, 42% of digital, growing 19% a year and forecast to reach INR 42,435 crore by 2027 [13].
Applying a 55% leak to programmatic spend suggests about INR 16,500 crore lost each year, rising to around INR 23,300 crore by 2027. Programmatic is only a proxy for the intermediated pool, since affiliate and direct-app deals aren't fully captured in it.
Looking ahead to 2030
If digital spend grows at around 7.9% a year, the rate dentsu forecast for 2025 [4], pure-play digital would reach about $1.16 trillion by 2030. Holding the leak at 26% puts it at roughly $303 billion a year.
That projection is directional, not a forecast. Growth rates vary year to year. But the leak share is more likely to rise than fall: fraud losses are projected to roughly double from $84 billion in 2023 to around $172 billion by 2028 [14][9], faster than the market itself.
Where this model could be wrong
- The 60/40 split is an operator estimate, not a published figure. It's the input that moves the answer most.
- The fee breakdown dates from 2020. The ANA's more recent tracker suggests the outcome still lands near 45 cents in open-web programmatic [15], but fees may have shifted within that.
- Leak rates vary by brand and channel. A brand with direct contracts and strict supply paths can do much better than average; one relying on resellers can do much worse.
- The model excludes softer waste. Poor targeting, weak creative and paying for users who would have converted anyway aren't counted.
The takeaway
The leak isn't a niche problem in a corner of the market. On conservative assumptions it's worth around $200 billion a year globally, about a quarter of all digital ad spend, and it's growing. That's the size of the prize for any approach that shortens the supply chain and pays for real outcomes rather than volume.
References
[1] WPP Media. This Year Next Year: Midyear 2025. June 2025. https://www.wppmedia.com/thought-leadership/this-year-next-year/tyny-midyear-2025-research
[2] WPP Media. This Year Next Year: December 2025. https://www.wppmedia.com/thought-leadership/this-year-next-year/December-2025
[3] WPP Media. This Year Next Year: Midyear 2026. June 2026. https://www.wppmedia.com/thought-leadership/this-year-next-year/midyear-2026
[4] 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
[5] 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
[6] AppsFlyer. State of Fraud for Marketers, 2026 edition. April 2026. https://www.appsflyer.com/resources/reports/state-fraud-marketers/
[7] AppsFlyer. Ad fraud data insights. April 2026. https://www.appsflyer.com/blog/measurement-analytics/ad-fraud-data-insights/
[8] mFilterIt. Reports index: Ad Fraud Report, Beyond the Linear Lens. December 2025. https://www.mfilterit.com/reports
[9] Search Engine Land (citing Juniper Research). June 2025. https://searchengineland.com/84-billion-lost-to-ad-fraud-what-you-missed-about-affiliate-abuse-in-2025-456530
[10] AppsFlyer. App marketing trends of 2025 (press release). December 2025. https://www.appsflyer.com/company/newsroom/pr/app-marketing-trends/
[11] exchange4media (dentsu-e4m 2026). Digital commands 59% of India's ad pie at INR 71,621 crore. February 2026. https://www.exchange4media.com/marketing-news/digital-commands-59-of-indias-ad-pie-at-71621-crore-dentsu-e4m-report-151570.html
[12] exchange4media (dentsu-e4m 2026). Indian advertising industry to grow at CAGR of 7.4% by 2027. February 2026. https://www.exchange4media.com/advertising-news/dentsu-e4m-report-indian-advertising-industry-to-grow-at-cagr-of-74-by-2027-151560.html
[13] exchange4media (dentsu-e4m 2026). Programmatic ad spends to grow at 19% CAGR by 2027. February 2026. https://www.exchange4media.com/digital-news/programmatic-ad-spends-to-grow-at-19-cagr-by-2027-dentsu-e4m-report-151566.html
[14] AppsFlyer. AI and ad fraud innovation. March 2025, updated November 2025. https://www.appsflyer.com/blog/mobile-fraud/ai-ad-fraud-innovation/
[15] ANA / TAG TrustNet. Programmatic Transparency Benchmark: Q1 2026. May 2026. https://www.ana.net/content/show/id/pr-2026-05-programmatic