Fraud filtering for affiliate trackers: Keitaro, Binom, RedTrack and Voluum
Score every click before it reaches the offer. What each tracker can do natively, where it stops, and the one-file gate that works with all four.
Keitaro has no VPN filter and its proxy check reads headers. Score each click with FraudNova from a custom PHP filter or a redirect gate.
Published
Keitaro runs on your own server, is written in PHP, and lets you add your own filters to a stream. That means you can score every click with FraudNova before Keitaro decides where to send it, and send proxy, VPN and bot traffic to a safe page instead of your offer.
This guide covers two ways to do it: a custom Keitaro filter that calls FraudNova directly, and a small redirect gate that works with any tracker. Both use the same rules and both cost 1 credit per new visitor IP per day. If you run more than one tracker, start with our overview of fraud filtering for affiliate trackers.
Keitaro ships with useful stream filters, but they were not built to tell you whether a visitor is hiding behind a VPN.
None of these gives you a single risk score for the IP, and none of them looks at the network the visitor is coming from. That is the gap FraudNova fills. A check on the visitor's IP returns a fraud_score from 0 to 100 plus flags such as proxy, vpn, tor and bot_status. You can see the full list on the proxy and VPN detection API page.
Keitaro loads custom filters from /var/www/keitaro/var/filters/. We publish a ready-made one: FraudNovaFilter.php. It needs no Composer and no extra libraries, only PHP's curl extension.
The heart of the file is short. Keitaro hands the filter the click and the value you typed into the filter, and the filter answers one question: is this click risky?
class fraudnova extends AbstractFilter
{
public function isPass(StreamFilter $filter, RawClick $rawClick)
{
$risky = self::isRisky(
(string) $rawClick->getIpString(),
$rawClick->getUserAgent(),
$rawClick->getLanguage(),
self::thresholdFrom($filter->getPayload())
);
return $filter->getMode() == StreamFilter::ACCEPT ? $risky : !$risky;
}
}
That is an excerpt; the full file adds the check itself. It sends the visitor's IP, User-Agent and language to GET /api/v1/check, and treats the click as risky when the fraud_score is at or above your threshold, or when the result says proxy, vpn, tor or bot_status. The value you type into the filter is the threshold: a whole number from 1 to 100, with 75 used when you leave it empty or type anything else.
Three details keep it safe to run on live traffic:
/var/www/keitaro/var/filters/fraudnova.php.fn_ and are created under Dashboard, API keys.In Reject mode a risky click does not pass the filter (the excerpt's last line), so it skips your offer stream and goes on to the safe-page stream below. Accept mode does the opposite: only risky clicks pass.
Keitaro's documentation explains how to write a custom filter, but it does not describe making outbound HTTP calls from inside one. The filter does exactly that, so a slow answer could hold a click for a moment. The 2-second timeout bounds that risk, the cache means most repeat clicks never wait at all, and the fail-open rule means a slow or failed check never blocks a visitor. Test it on a small stream first.
If you would rather not put code inside Keitaro, or you run more than one tracker, use the gate. It is a small script you host on your own server, in the tracker gate folder of our examples repo. There is a PHP version (php/gate.php, which needs PHP 8.2 and Composer) and a Node.js version.
The gate reads the visitor's real connection IP, User-Agent and Accept-Language from the request itself, never from a query parameter, so nobody can fake a clean verdict by adding ?ip= to the link. It checks the IP, then sends the visitor on with a 302 redirect to either your money page or your safe page. It uses the same rules as the filter: block at a fraud_score of 75 or above, or on proxy, vpn, tor or bot_status. You can change the threshold and the flags, turn on active_vpn, and limit traffic to a list of allowed_countries. It also caches each verdict per IP for 24 hours, waits at most 2 seconds for an answer, and lets the click through if the check fails.
In your offer stream, use the gate as the landing page or offer URL:
https://gate.example.com/gate.php?clickid={subid}&route=default
{subid} is Keitaro's own token: Keitaro replaces it with the click's id before the visitor arrives, and tokens like this work in any Keitaro URL. The gate receives that value as the clickid parameter and appends the incoming query string when it redirects, so clickid travels on to your money page. In the gate's own config, a {clickid} placeholder in a destination URL is filled with that same value.
Your money page stays as it was. Its offer link still goes back through Keitaro as usual, so conversions are attributed to the right click.
A blocked click never reaches your offer, so the gate can tell Keitaro about each one. Set report_url in the gate's config to your Keitaro postback, copied from Settings, Postback URL, with the status set to trash:
https://TRACKER/POSTBACK-KEY/postback?subid={clickid}&status=trash
The gate calls this once for every blocked click, with {clickid} filled in, and never for an allowed one. trash is one of Keitaro's standard statuses (so is rejected), so your reports can show how many clicks each source lost to the filter.
75 is where the high-risk band starts (0–39 is clean, 40–74 suspicious, 75–100 high risk), and it is a sensible start for most offers. If a bad click is expensive, for example on a finance or insurance offer, try 60 or 65. If you buy cheap, high-volume traffic and only want to cut the worst of it, try 85. The flags matter as much as the number: many affiliate networks reject proxy and VPN traffic whatever its score. Our guide to fraud score thresholds explains how to tune the number from your own results.
Every check is 1 credit, with no subscription. Because both the filter and the gate cache each verdict per IP for a day, you pay 1 credit per new visitor IP per day, not per click. New accounts get 25 free credits once the email is verified, which is enough to test a stream. Each API key has a rate limit and each account has a cap on checks running at the same time; both are listed in the API docs, and higher limits are available on request.
Take an IP from a recent click and run it through the free IP checker to see which signals come back. Then create an account for your API key, install the filter or the gate on one stream, and compare that stream's conversion rate with the rest of your campaign after a few days.
Run one IP proxy check right now — no account, no card.
Try the free IP proxy checker — no signup See the Proxy & VPN detection APIScore every click before it reaches the offer. What each tracker can do natively, where it stops, and the one-file gate that works with all four.
Voluum's Anti-Fraud Kit reports suspicious visits but cannot score a click against an API. Add a FraudNova gate to your path and block before the offer.
RedTrack filters proxy types and blacklists bots, but has no external scoring hook. Add a FraudNova redirect gate as your lander and block by score.