# Click Fraud Is Not Invalid Traffic. The Refund Depends on the Difference.

Source: https://bottrafficindex.com/blog/2026-11-11-click-fraud-vs-invalid-traffic-refunds.html

November 11, 2026

*The advertiser asks for a refund on 12,000 clicks. The platform refunds 3,000. The advertiser calls it theft. The platform calls it measurement. Both are using the wrong word, and the refund depends on the right one. Click fraud is not invalid traffic. Invalid traffic is the measurement category, and click fraud is one kind of traffic inside it, and the split between general and sophisticated invalid traffic is what decides how many clicks get refunded.*

## The definitions that decide money

The Media Rating Council defines invalid traffic as traffic that does not meet certain quality or completeness criteria, or that is not the result of genuine user interest. Inside that category, GIVT, general invalid traffic, is the traffic that can be identified through routine means: known crawlers, spiders, and non-browser user agents, and traffic from datacenter IPs. It is recoverable waste, in the sense that the platform filters it and the advertiser does not pay for it, or gets it back.

SIVT, sophisticated invalid traffic, is the traffic that needs advanced analytics to identify: bots that mimic human behavior, residential proxies, headless browsers, and click farms. It is fraud, not waste, because it is designed to look like a real user and to cost a real advertiser money. The distinction matters because the refund follows the classification: GIVT is filtered before billing or refunded routinely, and SIVT is detected after the fact, investigated, and refunded case by case, if at all. An advertiser who calls everything click fraud gets a fraud investigation. An advertiser who calls it invalid traffic, with the GIVT and SIVT split named, gets a refund process.

## GIVT vs SIVT in one paragraph

GIVT is the traffic a blocklist catches: a known crawler, a datacenter IP, a user agent that says it is a bot. SIVT is the traffic a blocklist misses: a residential proxy that looks like a home user, a headless browser that runs JavaScript and sets cookies, a bot that moves a mouse in a plausible pattern. The difference is not how much damage they do. It is how they are detected, and detection decides the refund, because a platform refunds what it can identify, and it can identify GIVT automatically and SIVT only with work.

The practical consequence is that the advertiser's evidence has to match the category. A GIVT claim needs the signatures: the IP ranges, the user agents, the timestamps. A SIVT claim needs the behavior: the mouse movements that repeat, the sessions that never convert, the clicks that arrive at regular intervals. A claim with neither is a complaint, and complaints do not get refunds.

## The refund claim, written right

Write the claim in the platform's language. Name the date range, the campaign, and the number of clicks. Split the claim into GIVT and SIVT, with the evidence for each: the datacenter ASNs and known crawler signatures for the GIVT share, and the behavioral patterns for the SIVT share. Ask for the GIVT refund first, because it is routine, and treat the SIVT share as an investigation, because it is. And keep the server logs, because the platform's measurement is not the only one, and the gap between the server logs and the analytics is the evidence that the traffic existed. For the taxonomy itself, see [GIVT vs SIVT and the MRC split](2026-11-10-givt-vs-sivt-mrc-invalid-traffic.md), and for the residential proxies that make SIVT hard to catch, [why the blocklist stopped working](2026-11-09-residential-proxies-why-blocklist-stopped-working.md).

## Frequently asked questions

### What is the difference between click fraud and invalid traffic?

Invalid traffic is the measurement category: traffic that is not genuine user interest. Click fraud is one kind of traffic inside it, the deliberate kind. Not all invalid traffic is fraud, and not all fraud is easy to prove.

### What is GIVT?

General invalid traffic: the traffic identifiable through routine means, like known crawlers, non-browser user agents, and datacenter IPs. It is filtered before billing or refunded routinely, because the platform can identify it automatically.

### What is SIVT?

Sophisticated invalid traffic: bots that mimic human behavior, residential proxies, headless browsers, and click farms. It needs advanced analytics to identify, and refunds for it are investigated case by case, because the platform has to prove the traffic was fake.

### What evidence does a refund claim need?

The date range, the campaign, and the click count, split into GIVT and SIVT. GIVT needs the signatures: IP ranges, user agents, timestamps. SIVT needs the behavior: repeating mouse movements, sessions that never convert, clicks at regular intervals.
