What Ad Fraud Rate Measures
Ad fraud rate is a simple ratio with a lot of judgement behind it. It tells you what share of the traffic you bought was identified as fraudulent. Most mobile teams track it at the install level: of all the installs attributed to a source in a period, how many did your fraud rules flag?
The formula is straightforward: flagged installs divided by attributed installs, times 100. The same idea applies to clicks (flagged clicks over total clicks) and to in-app events (flagged registrations or purchases over total). Each level answers a different question. Click fraud rate tells you about the quality of a network's inventory. Install fraud rate tells you about attribution. Event fraud rate tells you whether cost-per-action payouts are being gamed.
Why the Number Depends on Your Rules
A fraud rate is only as good as the rules that produce it. Two teams buying the same traffic can report very different rates because one checks click-to-install time, IP reputation and device patterns while the other only checks for data-center IPs. A rate that rises after you add a rule does not mean fraud got worse that week, only that you started counting it.
For this reason, avoid comparing your fraud rate with industry averages quoted without method. Compare your sources with each other under the same rules, and compare each source with itself over time. A sudden jump for one network, or a gradual climb for an affiliate, is a signal you can act on.
Reading Fraud Rate by Source
Break the rate down as far as your data allows: network, campaign, sub-publisher, country and platform. Aggregate fraud rates hide problems, because a clean high-volume source can mask a heavily fraudulent small one. Most fraud concentrates in a handful of sub-publishers, so the sub-publisher view is usually where action happens.
Look at the rule mix too. A source flagged mostly for click flooding needs different treatment from one flagged for SDK spoofing. Flooding points to attribution theft of organic installs; spoofing points to entirely fake users.
Weight by spend. An install fraud rate of 3 percent on a cheap source matters less than 3 percent on an expensive one. Spend-weighted fraud rate, the share of payouts going to flagged installs, is the number finance teams care about.
Tracking Fraud Rate in Linkrunner
Linkrunner's Fraud Control flags suspicious installs using templates and rules you write in plain English, covering click injection, install hijacking, SDK spoofing, click flooding by IP and by device, and installs from fraudulent clicks. Flagged installs stay in reporting totals, so you can calculate fraud rate per source directly from the same data. Each rule shows an impact preview before you turn it on, which helps you estimate how it will change the rate. Rules can tag installs or withhold their postbacks, so networks do not optimise toward flagged traffic. See fraud protection.
Acting on Fraud Rate
Set thresholds per source type. A self-attributing network and an open affiliate deserve different tolerances. When a source crosses its threshold, review the flagged installs, share the evidence with the partner and ask for the sub-publisher to be removed or for the payouts to be adjusted under your insertion order terms.
Keep withheld postbacks in place for flagged installs so ad algorithms train on real users. Revisit thresholds quarterly, and add rules as new patterns appear. The goal is not a fraud rate of zero, which usually means your rules are too narrow, but a rate you understand and a process that acts on it.
Fraud Rate and Other Quality Metrics
Fraud rate catches what you define as fraud. It does not catch low-quality traffic that passes every rule. Pair it with day-one and day-seven retention, revenue per install, and click-to-install time distribution per source. A source with a low fraud rate but very poor retention may be using a technique your rules do not yet cover.