Shadow 06 / Figure 8
Different samples, periods and units
Scroll the table sideways to see every column.
| Vendor | Reported figure | Measure type | Period | Unit | Sample | Extrapolation | Source |
|---|---|---|---|---|---|---|---|
| Spider Labs | 5.58% fraud rate; $25.3bn | rate (share of paid clicks) + extrapolated amount | H1 2026 | paid clicks | 4.006bn paid clicks | global half-year extrapolation | Spider Labs release, 2026 |
| Fraudlogix | 20.64% flagged invalid | rate (share of impressions flagged) | calendar 2025 | impressions | 105.7bn impressions | none stated | Fraudlogix, Ad Fraud Statistics 2026 |
| DoubleVerify | +140% CTV fraud schemes and variants | growth rate of a count (schemes and variants) | Q1 2026 vs Q1 2025 | identified schemes | vendor detections | none | DoubleVerify, 7 May 2026 |
| Thales/Imperva | 53% automated traffic; 40% bad bots | rate (share of observed traffic) | 2025 | web traffic | traffic observed by the vendor | none | Imperva Bad Bot Report 2026 |
| HUMAN | automated traffic grew 8x faster than human traffic | comparison of growth rates | 2025 | traffic growth rate | customer traffic | none | HUMAN, 2026 State of AI Traffic |
Vendors measure different things (rates of clicks, impressions or traffic; growth in a count of schemes; relative growth) over different periods and samples. The figures are not a range for one quantity.
SOURCE · Vendor publications named in the Source column; figures are vendor-reported.