A click-fraud percentage is only useful when you know what was counted. Global ad-fraud estimates, a vendor’s customer sample and your own campaign’s invalid-click report measure different things. This guide explains how to use published numbers without turning them into an unsupported claim about your budget.
A widely cited estimate, with its date attached
A Juniper Research report covering 2023–2028 estimated that 22% of global digital ad spend, or $84 billion, would be lost to ad fraud in 2023. Its published summary projected losses above $170 billion by 2028. These are broad market estimates and forecasts, not a measured 2026 Google Ads click-fraud rate. Read the report’s published summary.
The report is distributed by Fraud Blocker, an ad-fraud protection vendor, and attributed to Juniper Research. The year, scope and distinction between an estimate and a forecast are essential to interpreting it.
Numbers that should not share the same label
| Metric | What it describes | What to check |
|---|---|---|
| Global ad-fraud loss | An estimate across advertising markets | Year, channels, geography and methodology |
| Platform invalid clicks | Clicks a platform identified under its own criteria | Filtering and billing treatment |
| Vendor suspicious-traffic rate | Activity classified in that vendor’s observed sample | Sampling, thresholds and the denominator |
| Blocked sources | Sources against which an action was recorded | Successful external sync and block duration |
| Estimated spend exposure | A calculation from assumptions about activity and cost | Actual charges, credits and overlap |
Read the denominator before the percentage
A rate can use clicks, visits, sessions, people or impressions as its denominator. Repeated clicks by one visitor can therefore create very different percentages across reports. Changes in campaign mix, region or detection settings can also change the result without indicating a market-wide trend.
Before comparing vendors, ask for the observed period, channels, geography, sample selection, invalid-traffic definition and treatment of repeated activity. A precise-looking percentage without that context is not a useful benchmark.
Why suspicious visits are not the same as money lost
Google identifies and filters invalid activity through its own systems. Site-side detection is a different measurement, and not every suspicious visit is a billed ad click. Google’s invalid-traffic guidance explains the platform’s reporting context.
A calculation example, not a customer result
If 100 charged clicks cost $5 each, their gross cost is $500. That arithmetic does not establish that those clicks were fraudulent, that no credits apply, or that a protection tool would have prevented the full amount.
Build your own campaign baseline
- Choose a clear reporting period and align time zones.
- Separate campaigns, channels and the exact traffic metric.
- Record suspicious activity and the evidence supporting its classification.
- Check what the ad platform already filtered or credited.
- Track qualified conversions and the status of automatic protection actions.
Use AdProtektor to investigate your traffic
AdProtektor provides person-level traffic context, configurable automatic protection and evidence for reviewing decisions. Its Google Ads and Meta integrations use different exclusion mechanisms. Your traffic volume and campaign mix determine how much evidence is available during an evaluation.
Start a trial to build an account-specific view. Use published research to frame the question, then let your own evidence guide the decision.