Check a campaign's clicks for fraud

Before you pay an affiliate or renew an ad placement, you want one question answered: are these clicks real? The clicks report is a filterable table over every click your domains have ever logged — button taps, /r/<slug> smart-link hits, /t/<code> post links — with a fraud column computed honestly from what was actually captured, never guessed.

What counts as fraud here

A click is flagged fraud when either is true:

• the User-Agent captured at request time matches a script/crawler pattern (bot, crawler, curl, python, headless, …), or
• it's a repeat from the same coarsened /24 subnet + button on the same day (not is_unique) and a User-Agent was captured at all.

A click with no captured User-Agent (anything logged before edge-capture shipped) is never flagged — unknown is not the same as suspicious, and backdating history into "fraud" would be a lie about it.

Reading a campaign's numbers

1. Open the clicks report and narrow it to the campaign: filter by domain (exact match) and/or code (substring) to scope to one smart link or link inventory item, and by ad platform to isolate one traffic source (facebook, tiktok, google, telegram, and the rest are derived from the captured utm_source/referrer — an unrecognized source shows as blank, never guessed).
2. Add a period window to match the campaign's flight dates.
3. Read the totals row: clicks (everything that matched your filters), unique (first hit per subnet+button+day), converted (the click's id shows up in a real conversion), fraud.
4. Flip fraud only on to see just the flagged rows — click id, country, subnet, referrer host, and the UTM parameters that were captured, so you can trace exactly which rows are suspect and why.

A healthy campaign has fraud close to zero and unique close to clicks. A spike in fraud on one code or ad_platform points at exactly which placement to pause.

What you can't get from this report

click_id and external_click_id filters are substring matches — useful for tracing one visitor's journey, not for bulk fraud sweeps. And subnet is always the coarsened /24 — this report deliberately never stores or shows the full client IP.

What's next

Found the fraud but need to stop it at the source? See Hide content buttons from bots and ad reviewers.