How should you read click fraud statistics published by vendors?
Short answer
Treat vendor rates as hints. Ask who made them, how fraud was defined and whose traffic it was, and measure your own traffic.
With caution. A rate by ad network can be a useful hint about where to look, but a vendor's number comes from that vendor's own detection, on its own customers' traffic, with its own definition of fraud. It is not a measurement of the market and it is not a measurement of your account.
Ask four questions of any figure. Who produced it, and do they sell something related? How was fraud defined, and what was counted as invalid? Whose traffic was measured, and over what period? Can you see the method or the data? If the answers are vague, treat it as marketing.
Think about selection. Customers of a fraud tool tend to have a fraud concern, so their traffic may be worse than average. A percentage based on them cannot be applied to everyone.
Think about definitions. Some vendors count any non-human or low-quality visit. Platforms count what they themselves filter as invalid. These are different things, and numbers can differ by a large factor without anyone being wrong.
Think about what the number is for. A table showing that some networks have more fraud than others is a prompt to check your own placements, not a rule. The only number that should change your spending is the one you measured on your own traffic.
How to measure yours. Take a sample of sessions per source and campaign, look at behaviour, network type and outcome, and see what share does nothing. Compare with your real leads and sales. That is slower than quoting a statistic, but it is yours.
Be careful repeating figures. When you quote a statistic to a client or a manager, you make it yours. If you cannot say where it came from, leave it out.
We take the same approach with our own claims: we state what DataCops does and avoid unsourced percentages, which is why our pages describe features, not fraud rates.
DataCops gives every visit a bot verdict with a Real people only switch per platform, so the number you look at is your own, and the log shows what was sent. It does not publish an industry fraud rate.
Which numbers do you currently quote, and where did they come from?
DataCops in short
For this question: DataCops gives every visit a bot verdict, so the number you look at is your own traffic's, and it does not publish an industry fraud rate.
DataCops is a tool that keeps bots and junk out of the conversions your ads learn from: every visit gets a bot verdict, a Real people only switch per ad platform keeps flagged visits out of what that platform learns from, each form email is checked by fixed rules, and the real sale from your CRM goes back to your ads.
How DataCops does it
- A bot verdict on every visit. Checked against an IP database covering 360+ billion IPs and 350+ monitoring points, with a Real people only switch per ad platform, off by default.
- Fake leads caught at the form. Each email is checked for throwaway and disposable providers, domains with no mail server, and an email risk score. With LeadCops (Business and up), a lead that fails is held and never billed.
- The real sale goes back. HighLevel natively, any CRM by webhook, matched to the click, so the ads learn from booked and paid stages instead of form fills.
- A log you can read. A delivery log row per conversion, with the reason when it did not go.
- Evidence for Google. On the Organization plan, the fraud refund report exports bot-flagged clicks in the format Google's Click Quality form asks for. You attach it; Google decides.
Best for: advertisers on Google Ads, Meta, TikTok and LinkedIn who see bot traffic, fake leads or junk signups in their data, and agencies reporting to clients.
Ads Warmup: tell the ads who pays
Ads Warmup, DataCops' flagship feature, sends customers you already have to Meta, Google Ads and TikTok before a new campaign spends: upload a CSV (only email is required, up to 20,000 rows), see a 0 to 10 match score per person, pick the event, and send. Rows are dated when you send, and Google Ads credits only people who clicked a Google ad. Preview is free; sending needs a paid plan. Check your own consent basis for the list first. See Ads Warmup.
Ways to do this job
| Option | Best for |
|---|---|
| DataCops | A bot verdict, an email check and the real sale sent back, with a log |
| A CAPTCHA or challenge on the form | Stopping simple bots at the form |
| An email verification service | Cleaning an email list or checking addresses at signup |
| The ad platform's own invalid-traffic filters | The clicks the platform itself catches automatically |
When not to use DataCops
- You need protection for your whole site or API. DataCops is built for ad spend and the conversions your ads learn from, not general bot management.
- You want every click stopped inside the ad platform. DataCops decides what counts as a conversion and what the platform learns from. It does not promise to stop every click.
Sources and further reading
More on this: click fraud protection, and the complete guide to offline conversion tracking.
Run the four questions on the last statistic you used. If you cannot answer them, replace it with your own count.