The invalid traffic tax

Thirty-five cents of every dollar spent on open-web programmatic advertising in 2023 went to inventory that never reached a human being — invalid traffic, made-for-advertising farms, or impressions nobody could see or measure — according to the Association of National Advertisers’ audit of $123 million in real campaign spend. Ask a verification vendor the same question about its own protected inventory and the number drops more than fifty-fold, to under one percent. Both figures are defensible. They measure different things, and the gap between them is the real story of ad fraud in 2026 — not one number, but a dispute over what counts as “invalid” and at which point in the pipe you measure it.

Four scorecards, four different games

Four organisations that track invalid traffic publish four rates that do not sit on the same scale at all — from a fraction of one percent to more than a third of spend. The difference is not precision; it is scope.

Why a fraud rate depends on which side of the filter you stand

The cleanest proof that methodology drives the headline sits inside IAS’s own report: the same report, the same year — switch on fraud-mitigation technology and the fraud rate drops from 10.9% to 0.7%, a fifteen-fold gap produced entirely by which inventory a campaign was allowed to touch. DoubleVerify’s 0.5% and IAS’s 0.7% describe what survives after a verification vendor has already blocked the obvious junk; the ANA’s 35% describes what a marketer’s dollar meets before any such filter is applied, across an average of 44,000 sites and apps per campaign. Ask “how much invalid traffic is there” and the honest answer is: in which layer of the pipe?

The fastest-growing slice: made-for-advertising

Made-for-advertising sites — pages engineered to maximise ad slots per pageview rather than to inform anyone — made up roughly 5% of open-web ad auctions in early 2020 and nearly 30% by mid-2023, per the ANA study. Video inventory is not exempt from the shift toward newer, less-audited formats: DoubleVerify recorded 140% more CTV fraud schemes and variants in the first quarter of 2026 than a year earlier, in a market where unprotected inventory can cost advertisers roughly $1.8 million in fraud losses per billion CTV impressions served.

A bill that keeps climbing even as protected rates fall

The apparent paradox: verification vendors report their own protected-side numbers improving — DoubleVerify’s violation rate fell 24% year-on-year to 0.5% in the first quarter of 2026 — while economy-wide estimates keep rising. Juniper Research put global ad fraud losses at $59bn in 2021 and $68bn in 2022, then $84bn in 2023 (22% of that year’s $382bn in global online ad spend), and forecast the total would reach $172bn by 2028, a projected 23% share. The two trends are not necessarily contradictory: total programmatic spend keeps growing, and fraud migrates toward whichever channel is newest and least audited — the open web in 2023, connected TV in 2026.

Methodology

Sources

ANA, “Programmatic Media Supply Chain Transparency Study,” published Dec 5, 2023 — audit of $123 million in tracked ad spend and 35.5 billion impressions across 21 marketers and 12 supply-chain companies, September 2022–January 2023, via martech.org (https://martech.org/ana-study-finds-25-of-programmatic-ad-dollars-are-wasted/). Integral Ad Science, “Media Quality Report,” 20th edition, published May 24, 2025, covering 2024 global data across 280bn+ daily digital interactions, via mediabrief.com (https://mediabrief.com/ias-media-quality-report-ad-fraud-2025/). DoubleVerify, Quarterly Benchmarks report for Q1 2026, covering 100bn+ monthly indexed impressions, via ppc.land (https://ppc.land/doubleverify-cuts-fraud-rate-24-but-out-of-geo-ads-jump-12/); DoubleVerify, “Must-CTV: Streaming’s Shift From Promise to Performance,” published May 7, 2026 (https://doubleverify.com/company/newsroom/global-study-ctv-fraud-schemes-surge-140-globally). Juniper Research figures for 2023 and the 2028 forecast via Marketing Brew, Oct 5, 2023 (https://www.marketingbrew.com/stories/2023/10/05/advertisers-will-spend-usd84-billion-on-ad-fraud-this-year-report); 2021–2022 figures via Juniper Research press release, Feb 21, 2022 (https://www.juniperresearch.com/press/digital-advertising-spend-lost-to-fraud-68-billion/).

Scope limits

These four figures are not directly comparable and this brief does not collapse them into one number. The ANA figure covers open-web display and video programmatic only, sampled from large mostly-US marketers, before verification filtering is applied. The IAS and DoubleVerify figures describe post-filter, vendor-protected inventory, self-reported by the vendors that sell the filtering technology, and are not independently audited outside their own published reports. Juniper Research’s totals are a modelled market forecast rather than a sampled audit, and its historical figures have been revised between report editions — the $68bn 2022 estimate and the $84bn 2023 estimate come from separate reports, not one continuous series. All figures are global aggregates; none isolate UK or EU markets specifically.

Updates

This brief will be revised as new data is published. Corrections are handled under our corrections policy.