GIVT vs SIVT: what is the difference?
GIVT (general invalid traffic) is ad traffic that routine filters can identify: declared crawlers, known data-center IP ranges, non-browser user agents and pre-fetched pages. SIVT (sophisticated invalid traffic) takes advanced analysis to find: bots posing as people, hijacked devices, adware and malware, and spoofed sites or apps. The Media Rating Council defines both classes.
Updated
Where the terms come from
The Media Rating Council, which accredits ad measurement in the United States, split invalid traffic into two classes in its invalid traffic detection and filtration guidelines, first published in 2015. The split is about how hard the traffic is to detect, not how harmful it is. Measurement the MRC accredits filters GIVT as a baseline; SIVT detection takes more work and is accredited on its own, so vendors state whether theirs covers it.
Two shared lists support the routine part: the IAB Tech Lab Spiders and Bots List of declared crawlers, and the Trustworthy Accountability Group's Data Center IP List of hosting ranges that send non-human traffic.
What GIVT covers
GIVT is traffic that announces itself or comes from places that consistently produce non-human traffic. Most of it is not fraud: a search engine crawler loading a page with ads on it is GIVT.
- Declared crawlers, spiders and monitoring bots, identified by user agent or list.
- Known data-center traffic: hosting and cloud ranges that are a consistent source of non-human requests.
- Non-browser user agents and unknown browsers.
- Pre-fetch and pre-render loads where no person ever saw the ad.
- Simple activity rules, such as implausible repeat counts from one source in a short window.
What SIVT covers
SIVT hides. Finding it takes analysis across many signals and sessions, sometimes with human review.
- Bots and automated browsers that present themselves as real users.
- Hijacked devices and sessions, and adware or malware that loads or clicks ads without the user knowing.
- Domain and app spoofing, where a bid request claims to come from a site or app it does not.
- Incentivized or manipulated activity, such as paid clicks, or falsified viewability and location data.
- Cookie stuffing and other attribution fraud.
GIVT and SIVT compared
| Aspect | GIVT | SIVT |
|---|---|---|
| Found by | Lists and standard checks | Analysis across many signals, models and human review |
| Typical sources | Crawlers, data-center IPs, pre-fetch | Disguised bots, hijacked devices, malware, spoofed inventory |
| Usual intent | Often harmless | Usually deliberate fraud |
| Filtering | Baseline in accredited measurement | Accredited separately; check the vendor's scope |
What it means for ad verification checks
A verification check is non-human traffic by definition, so it falls into one of these classes. From a datacenter IP it is GIVT: many ad stacks filter it before the auction, so the check sees fewer ads, different ads or a clean landing page. From a residential IPv4 exit the data-center rule does not apply, but SIVT models still read the browser, the behavior and the exit's history, and heavily reused proxy exits can be flagged.
Keep checks small and passive. Load pages and never click ads: a click from a check is an invalid click on someone's budget. Record the exit, the time and the response so a reviewer can tell your sample apart from the traffic being measured. Ad verification tools vs proxies explains where vendor measurement ends and spot checks begin.
Reading an invalid traffic report
- Ask whether the rate covers GIVT only, or GIVT and SIVT. Reports often show them separately.
- Ask whether traffic blocked before the bid is counted. Pre-bid filtering lowers the post-bid rate without changing how much invalid traffic was offered.
- Compare rates by placement and app, not only by campaign. Spoofed and hijacked inventory clusters in a few sources.
- Check what the ad platform credits back. Platforms generally do not charge for invalid traffic they detect, but the rules differ by platform and campaign type.
GIVT, SIVT and TrueProxies exits
Residential IPv4 exits are on residential ISP networks, so they do not fall under the data-center rule for GIVT; SIVT scoring still applies to everything else about a check. Datacenter IPv6 is hosting traffic and is not recommended for ad checks. The ad verification proxies page shows which product fits which check, with blank evidence templates.
Questions about GIVT and SIVT
Is GIVT fraud?
Usually not. GIVT is mostly declared crawlers, monitoring and pre-fetch that no person saw. It is removed from measurement and billing because it is not an audience, not because someone meant harm. Deliberate fraud usually falls under SIVT.
Is all data-center traffic invalid?
No. The lists aim at hosting ranges that consistently produce non-human traffic, and some people do browse through data centers, for example behind corporate gateways or VPNs. In practice many ad stacks still treat data-center addresses as GIVT, so a check from one rarely sees what a household sees.
Can residential proxy traffic be SIVT?
Yes. Network type is one input. SIVT models also read device and browser signals, behavior and an address's history, and traffic from heavily reused proxy exits can be classified as sophisticated invalid traffic.
Who decides what counts as invalid traffic?
The Media Rating Council sets the measurement guidelines, with shared lists from IAB Tech Lab and the Trustworthy Accountability Group. Each ad platform and verification vendor applies its own detection within that framework, so two vendors can report different rates for the same campaign.