Click fraud
Click fraud in the Marketing area (Attribution) — route /marketing/fraud.
Click fraud lives in the Marketing area of the dashboard, under Attribution.
At a glance#
| Dashboard route | /marketing/fraud |
| Area | Marketing (marketing) |
| Group | Attribution |
| Platforms | Available for every app platform. |
What it does#
Click-fraud detection. It reports invalid traffic from materialised rollups — a summary, breakdowns by channel, campaign and source, trends, and the reasons clicks were flagged — plus a feed of individual suspicious clicks with IP and fingerprint detail you can investigate.
Its job is to catch invalid traffic before it pollutes your ROAS. Fraudulent clicks that get attributed inflate a channel's apparent performance, which makes you spend more on it — the mechanism is self-reinforcing, which is why it is worth a dedicated page.
When to use it#
Whenever a channel performs suspiciously well on installs and suspiciously badly on everything afterwards. That signature — high volume, low activation, no revenue — is what click fraud looks like from your side.
Also before scaling any new source, especially a smaller network or an affiliate. Fraud concentrates in the long tail.
Workflow#
Read the by-source breakdown first
Fraud is rarely spread evenly. One source usually accounts for most of it.
Check the reason breakdown
Different reasons mean different things — a spike in one reason code is a specific technique, and it tells you what you are dealing with.
Investigate the click feed for concentration
Many clicks from one IP or one device fingerprint is the clearest possible signal. The feed is where you confirm it.
Take it to the partner with evidence
The IP and fingerprint detail is what makes a chargeback conversation possible. A suspicion is not; a pattern is.
Permissions and prerequisites#
Requires attribution to be running, since fraud detection operates on click and install traffic.
Limits and edge cases#
Rollups are materialised, so figures reflect the last computation rather than the current second.
Flagged is not excluded. Check whether flagged traffic is still counted in your attributed totals before assuming your ROAS is already clean.
Fraud techniques change. Detection that worked last quarter may not this one.
Troubleshooting#
High fraud on one source. Investigate the click feed, then raise it with the partner. Pause the source while you do.
Fraud flagged on our own organic traffic. Look for a shared-network cluster — an office or a testing environment generating many clicks from one address.
No fraud detected at all. Either genuinely clean traffic, or you are not running the paid channels where it occurs.
Where the data comes from#
Served by
Fraud