Invalid traffic (IVT) is any ad activity that does not come from a real person with a genuine chance of seeing the ad. It wastes advertiser budgets, and it also hurts honest publishers through clawbacks, blocked accounts and lower bids. This guide explains the main categories of IVT, why no single check catches everything, and what publishers should watch in their own data.
GIVT and SIVT: two very different problems
The Media Rating Council (MRC), the US body that accredits ad measurement, splits invalid traffic into two categories. The distinction matters because each needs a different kind of defense.
General invalid traffic (GIVT) is the easy-to-identify kind. It includes:
- Known search engine crawlers and other declared bots
- Traffic from known data center IP ranges
- Automated activity that identifies itself, such as monitoring tools
- Obviously irregular patterns, such as repeated requests from the same source in a way no person would produce
GIVT can usually be filtered with lists and simple rules. Much of it is not malicious; a search crawler is doing its job, it just should not be counted as an audience.
Sophisticated invalid traffic (SIVT) is deliberately designed to look human. It includes:
- Bots that mimic mouse movement, scrolling and clicks
- Hijacked devices and malware generating hidden ad calls
- Domain and app spoofing, where low-quality inventory is labeled as a premium site or app
- Ad stacking and pixel stuffing, where ads are layered or shrunk so they cannot be seen
- Incentivized or manipulated behavior that misrepresents real engagement
SIVT requires deeper analysis: behavioral signals, device checks, cross-referencing across many requests and continuous model updates. Fraudsters adapt, so the detection has to adapt too.
Why one layer is never enough
No single check sees everything. A blocklist catches known bad actors but not new ones. A behavioral model catches patterns but needs data to accumulate first. A post-campaign report catches problems after the money is spent.
Layered protection combines checks at different points in the transaction so that what one layer misses, another can catch. It also spreads responsibility: publishers, supply platforms, marketplaces and buyers each see different signals, and each is in the best position to act on some of them.
Pre-bid and post-bid checks
The two main timing layers are before the auction and after the impression.
Pre-bid
Pre-bid checks evaluate a bid request before any buyer spends money on it. They typically look at:
- Whether the IP, device or user agent is on a known invalid list
- Whether the domain or app bundle ID in the request matches the real source
- Whether the seller is authorized in the publisher’s ads.txt or app-ads.txt
- Whether the request carries a complete, verifiable supply chain
Pre-bid filtering is efficient because it prevents waste instead of refunding it. Its limit is that it only sees the request, not what happens after the ad loads.
Post-bid
Post-bid checks run after the ad is served, using measurement tags or server-side signals. They can see whether the ad rendered, whether it was viewable, how the device behaved and whether engagement looked human.
Post-bid data catches sophisticated patterns that only show up in behavior. It also feeds back into pre-bid rules: a source flagged post-bid can be blocked pre-bid next time. The two layers are strongest when they share information.
Signals publishers should monitor
Publishers are often the last to hear about IVT, typically through a partner report or a deduction. Watching your own data lets you catch problems earlier and protect your reputation with demand partners.
Keep an eye on these signals:
- Traffic source changes. A sudden rise in visits from one referrer, campaign or country, especially from paid traffic acquisition, deserves a closer look. Bought traffic is a common entry point for bots.
- Engagement that does not fit. Very short sessions with high pageviews, near-zero scroll depth or perfectly regular timing between actions suggest automation.
- Click-through rates out of line. Clicks that spike without a matching change in content or placement can indicate click fraud or accidental-click layouts.
- Data center and unusual device mixes. A growing share of traffic from server IP ranges or outdated browser versions is a common GIVT marker.
- Discrepancies between partners. Large gaps between your ad server counts and a partner’s counts can point to filtering on their side. Ask what was removed and why.
- Unauthorized sellers. Periodically check that the partners selling your inventory match your ads.txt, and watch for reports of your domain being spoofed elsewhere.
What to do when something looks wrong
Pause the suspect traffic source first, then investigate. Share findings with your demand partners proactively. A publisher who flags and fixes a problem is treated very differently from one who waits for a clawback.
How buyers benefit from the same checks
Advertisers want the same thing publishers want: real people seeing real ads. Layered checks give buyers confidence that curated supply is what it claims to be, which supports stronger demand for clean inventory. Buyers should still run their own verification, but supply that is filtered at the source makes those reports cleaner and easier to act on.
Key takeaways
- GIVT is easy to identify and filter with lists and rules; SIVT is designed to look human and needs behavioral analysis.
- No single check catches everything; pre-bid filtering prevents waste, and post-bid measurement catches what only shows up in behavior.
- The layers work best when post-bid findings feed back into pre-bid rules.
- Publishers should monitor traffic sources, engagement patterns, device mix, partner discrepancies and ads.txt authorization.
- Acting early and sharing findings protects a publisher’s standing with demand partners.
Fraud protection works best when it is built into the marketplace rather than bolted on afterward. Vortex Media applies multi-layer fraud checks and creative filtering across curated web, in-app and CTV supply, so both publishers and advertisers trade with more confidence. To talk through your setup, contact our team.
