Free affiliate fraud scan, in your browser.

This free affiliate fraud scan reads a Rewardful, FirstPromoter, Tapfiliate or Tolt export in your browser. It replays your history sale by sale through our fraud rules and shows who the engine would have held, and for how much.

Runs in your browserNo signup, no uploadThe engine's own thresholds

The scan runs on your computer. Your file is not sent to us or anyone else. Close the tab and it is gone.

Your browser reads the file. The scan never uploads it.

The same rules that hold live sales

The scan replays your history in time order. At each sale it uses only the sales before it, the way the engine screens a live sale. The thresholds are the engine's.

Engine rules a sales export can run

  • Self-referral. The buyer's email matches the affiliate's after alias cleanup. See how the own-link email match works.
  • Shared buyers. 2 or more of an affiliate's buyers also bought via another affiliate. See the ring rule and its test.
  • Coupon hijack. 3 or more sales on one coupon in 7 days, taking customers first seen with another affiliate. See leaked coupons, tested at 2 and 3.
  • Sales bursts. 30 or more sales for one affiliate in an hour.
  • Odd hours. 20 or more sales in 7 days, over half between 00:00 and 05:59 UTC, and z above 3 against the program.
  • Click speed and shared IPs. Only when the export has an IP column. The scan uses the sale's IP where the engine uses the click's.
  • Duplicate accounts. The one rule an affiliate-list export can run.

Two checks the engine doesn't have

  • Refund rate. 5 or more sales with 20% or more refunded or charged back.
  • Sale seconds after the click. 3 or more sales under 10 seconds after their click.

Those two thresholds are our defaults, not industry figures. Results mark them as scan-only.

What a scan can't see

An export is history, not live traffic. Bot, datacenter, geo and IP-fallback rules need data exports don't carry, so the scan lists them as not run, with the reason. It never looks IPs up online.

The column names are our best reading of each tool's exports. We haven't yet tested them on a live export from every tool. When a rule's columns are missing, the scan skips it and says why.

Why give this away?

Because the scan is the product's argument. If your program is clean, you should know that for free. If it isn't, you should know what the money rules would have caught before you paid it out.

AffiliateTracking runs these rules on every sale, holds flagged commissions before payout, and shows you the evidence to release or uphold each one. See the full engine.

Our data

The scan and the engine agree

  • 8 synthetic exports, 2 per tool, 151 rows in all.
  • The scan flagged 72 sales across the 4 sales exports, from 10 different rules and checks.
  • A Go test loads the same rows into the real engine one by one. For the engine's rules, its per-rule, per-affiliate counts match the scan's.

Both test suites passed on : node tools/fraudscan/test.mjs and go test ./tools/fraudscan/.

What fraud might be costing you

A quick guess, on numbers you type. It stays on this page, like the scan.

    The presets (2%, 5%, 10%) are round starting points for your guess, not industry figures. Nobody has honest industry numbers for this.

    Scan questions

    Does my file get uploaded?

    No. The scan is two small scripts that read the file on your computer and make no network calls. A test fails if either script tries to send anything.

    Which export should I pick?

    A sales list: referrals, commissions or conversions. That is where the real checks run. An affiliate-list export can only check for duplicate accounts, because it has no sales in it.

    What does a finding mean?

    It means the live engine, seeing the same history, would have put those sales on hold before payout and shown you the evidence. You decide whether to release or uphold. A finding is a lead to check, not a verdict. The two scan-only checks are our own flags, not engine holds.

    Does the scan change anything?

    No. It reads your file and shows results. Your tracker, your file and your data are untouched.

    Why do some rules say not run?

    Each rule lists what it needs, like sale timestamps or IP columns. If your export's columns don't carry that, the rule is skipped with the reason shown, never guessed at.

    Found something?
    Hold it before you pay it.

    AffiliateTracking runs these rules on every live sale, with the evidence to decide. Self-host for free, or let us run it for you.