Digital marketers keep quoting one number: $63 billion lost to invalid traffic. Dr. Augustine Fou, the ad fraud researcher behind FouAnalytics, says that number is built on the wrong layer of measurement. In a column published October 16, 2026 on State of Streaming, he argues that click-based invalid traffic rates miss the bulk of impression fraud on CPM display and video buys.
Clicks are not impressions
Fou’s core point is simple. In a CPM campaign, bots only need an ad to load to get paid; they have no reason to click. In a CPC campaign, the incentive reverses. So a website’s own count of fake visitors says little about how many display impressions never reached a person.
The evidence is a pair of FouAnalytics charts from an unnamed advertiser, covering January 8 to February 7 of an unstated year. The impression chart shows the dark red invalid band climbing to roughly 80% to 85% in the later weeks. The on-site chart stays mostly below 10% for dark red invalid visits. The campaign, exchanges and total impressions are not disclosed, and the on-site chart covers all site traffic, not just those display buys.
What the $63bn figure actually measures
Fou does not name the study, but the details match Lunio’s 2026 Global Invalid Traffic Report. Lunio analysed more than 2.7 billion paid ad clicks from August 1, 2024 to July 31, 2025 across Google, Meta, Microsoft Bing, LinkedIn, TikTok and X, and found 8.51% invalid traffic.
- TikTok: 24.2%
- LinkedIn: 19.88%
- X: 12.79%
- Microsoft Bing: 10.32%
- Meta: 8.2%
- Google Ads: 7.57%
Within Google, Display landed at 12.02% and Video at 20.62%, against 5.21% for Search. The report then applies that rate to global digital spend to reach $63 billion – a calculation Fou says should not be made with click data for CPM-heavy inventory. Note that Lunio also sells detection, and FouAnalytics is a competing verification tool, so neither side is neutral.
IP addresses no longer sort bots from humans
Fou’s second objection is that detecting bots by IP address fails in both directions. Botnets route traffic through residential proxies, often built from free VPN apps on compromised home devices, so they look like real households. Meanwhile, humans using VPNs appear on data center IPs and get blocked. His recommendation is to stop filtering invalid clicks out of site reports, because blocking them hides the problem instead of fixing the campaign.
What buyers should do
For performance marketers, click-based IVT rates are usually the only fraud number that reaches the budget conversation. That is exactly why the gap matters. Ask verification vendors and platforms for impression-level invalid traffic on display and video buys. Compare that with the site-side click rate before deciding how clean a channel is. Keep invalid clicks visible in analytics so problems like Audience Network placements can be spotted. But do not treat one undisclosed campaign’s 80% figure as a market average: Fou himself offers no global estimate and argues that big scary numbers rarely change behaviour anyway.
Source: PPC Land
