What is invalid traffic, and what does a low IVT rate really prove?
Part of MFA avoidance
The short answer
Invalid traffic (IVT) is ad and content activity that should not count in measurement, such as bots, data center traffic, hidden ads or manipulated clicks. The Media Rating Council splits it into general IVT, caught with lists and routine checks, and sophisticated IVT, which needs advanced analysis. Accredited measurers must remove known general IVT from monetized counts.
Key takeaways
- The Media Rating Council defines invalid traffic broadly: activity that does not represent legitimate traffic that should be included in measurement counts.
- General invalid traffic is found with lists and parameter checks; sophisticated invalid traffic needs advanced analytics, corroboration across sources and often human review.
- Known general invalid traffic must be removed from monetized counts, while sophisticated invalid traffic may be removed later and must be reported separately.
- A site can show low invalid traffic and still be made for advertising, because its visitors can be real people who arrived through bought links.
- Ask every measurement partner for its decision rate, since impressions it could not judge are reported as unknown rather than valid.
What is invalid traffic in advertising?
The Media Rating Council (opens in a new tab) (MRC) defines invalid traffic generally as traffic or associated media activity that does not meet certain quality or completeness criteria, or otherwise does not represent legitimate traffic that should be included in measurement counts. The definition covers impressions and also the metrics built on them, such as viewability, clicks, engagement and outcomes.
Definition
- Invalid traffic (IVT)
- Ad or content activity that a measurer removes or segregates from reported counts because it came from non-human sources or from activity designed to produce invalid counts. The MRC uses IVT as the collective term for its general and sophisticated categories.
Invalid traffic is a measurement term, not a legal one. An impression can be invalid because a search engine crawler loaded the page, which is nobody's fault, or because someone stacked ads in one slot to bill for them, which is deliberate. Routine data cleaning, such as removing corrupt records or duplicates, is not invalid traffic unless evidence points to one of the listed techniques.
How do general and sophisticated invalid traffic differ?
The MRC's June 2020 update, known as IVT 2.0, sorts invalid traffic into two categories by how hard each is to detect. The split matters for buyers because the two categories carry different obligations: every accredited measurer must apply general detection, while sophisticated detection is strongly encouraged and accredited on its own. Our field guide to the types of ad fraud explains how the individual schemes work.
| Aspect | General invalid traffic (GIVT) | Sophisticated invalid traffic (SIVT) |
|---|---|---|
| How it is found | Routine filtration through lists or other standardized parameter checks | Advanced analytics, multi point corroboration and significant human intervention |
| Examples in the standards | Known invalid data center traffic; declared bots, spiders and crawlers; non-browser user agents; pre-fetch traffic; 0x0 and 1x1 ad sizes; sessions that cannot render images | Hijacked or infected devices; click farms; forced clicks and clickjacking; falsified impressions or locations; domain laundering and app ID spoofing; stacked or hidden ads; ad injection by adware |
| Who must apply it | Every accredited or certified measurement organization | Strongly encouraged; eligible for separate accreditation |
| When it leaves the counts | Known GIVT must be removed from monetized counts | May be removed later to protect detection methods from reverse engineering, and must be segregated when net metrics are reported |
| Typical buyer question | Does the vendor apply the industry lists and its own data center rules? | Which SIVT categories does the vendor's accreditation cover? |
The line between the two moves over time. The standards expect that some SIVT techniques will become standardized, objective lists and then be reclassified as GIVT. An April 2024 MRC interim update (opens in a new tab) did something similar for data center traffic: it noted that industry lists have limits, and began requiring filtration of invalid traffic from IP addresses of the three largest known hosting providers as a starting point. The same memo addressed privacy limits on IP and user agent data, domain and bundle ID mismatches in connected TV, and property level considerations.
Who defines invalid traffic standards?
In the United States the reference is the MRC, which issued its Invalid Traffic Detection and Filtration Guidelines in October 2015 and replaced them with the IVT 2.0 standards addendum in June 2020. The addendum was created with the IAB, the MMA and measurement practitioners, and reviewed by the 4A's and the ANA. The MRC lists both documents on its standards and guidelines page (opens in a new tab). Accreditation is granted after an independent audit of a measurement service.
| Body | Role in invalid traffic | What a claim citing it can and cannot show |
|---|---|---|
| Media Rating Council | Writes the IVT detection and filtration standards and accredits measurement services after audits | Shows that a measurer's methods were audited against the standards; says nothing about a single campaign's results |
| TAG (Trustworthy Accountability Group) | Runs the Certified Against Fraud (opens in a new tab) program for companies that abide by its guidelines, and maintains an IVT taxonomy | Shows a company's participation and compliance with TAG guidelines; it is a certification of the company, not of every impression |
| IAB Tech Lab and the Alliance for Audited Media | The IAB/ABC International Spiders and Bots List, maintained by AAM for the Tech Lab, which the MRC names as an input to general filtration | Shows that declared crawlers can be filtered consistently; it does not catch bots that hide what they are |
How is invalid traffic filtered from billing?
Filtering happens at several points, and a buyer sees the results in different places. Before the auction, a buying platform can decline requests that match its own lists. At serving and measurement, the MRC standards require known GIVT to be removed from monetized counts, and GIVT to be excluded where possible from processes that affect monetization, such as targeting and frequency capping. SIVT can take longer to detect, so it may be removed after the original report.
Google's help page on managing invalid traffic (opens in a new tab) describes the billing side for Google Ads. Invalid traffic, including impressions, clicks, interactions and conversions, detected before billing is not charged and is removed from billing and reports. Activity found after billing is credited as an invalid activity adjustment where appropriate and possible, and advertisers can request an investigation for activity within the last 60 days. Google lists examples such as bots, accidental clicks from poor ad placement, clickjacking and ad stacking.
Reconciling invalid traffic after a flight
- Pull the platform's invalid activity figures. Note which were removed before billing and which came back as credits.
- Pull the verification partner's report for the same dates. Record its GIVT and SIVT rates separately, and its decision rate.
- Line up the two by domain or app. Large gaps on a few placements are more useful than a small gap across the whole buy.
- Check contract terms for SIVT found later. Agree in advance whether late findings lead to credits, make goods or exclusions.
What does an unknown or undecided impression mean in an IVT report?
IVT rates are only as good as the share of traffic the measurer could judge. The MRC standards require a decision rate: impressions without enough signals for an IVT determination must be reported as unknown, not assumed valid or invalid. The standards give a worked case of 100 rendered impressions, where 85 could be judged and 15 could not, for a decision rate of 85 percent. Only the 85 judged impressions feed the IVT rate and downstream metrics such as viewability.
For a buyer this means a 1 percent IVT rate on a 60 percent decision rate describes a smaller slice of the buy than the same rate on a 98 percent decision rate. The standards also ask for the decision rate by environment, such as desktop and web, in-app and OTT, because tags and signals differ across them. Ask for the decision rate next to every invalid traffic figure in a report.
Why can a site with low invalid traffic still be made for advertising?
Invalid traffic asks whether an impression was real. Made for advertising asks why the page exists. When the ANA, the 4A's, the WFA and ISBA defined MFA in September 2023, the ANA's announcement (opens in a new tab) said MFA websites exhibit low levels of invalid traffic. The visitors can be real people who clicked a paid link, so they pass a fraud filter. Our MFA explainer covers the five MFA traits, and the traffic arbitrage guide covers how paid visits are bought and resold.
Purchased traffic is where the two topics touch. IVT 2.0 defines purchased traffic as visits redirected from properties the acquiring entity does not own, or incentivized activity. It requires measurers to report purchased traffic separately where known, says undeclared purchased traffic should be considered for filtration, and says obfuscated purchased traffic or purchased traffic with robotic characteristics shall be filtered as invalid. Declared purchased traffic from real people can stay valid, which is how an arbitrage site keeps a clean IVT report.
| Check | Question it answers | What it misses |
|---|---|---|
| Invalid traffic rate | Were these impressions served to real people on real devices? | Whether the people came through bought links, and whether the page was made mainly for ads |
| Viewability | Was the ad on screen long enough to count? | Ad clutter and refresh, which can raise viewable impressions per visit |
| Brand safety | Did the ad appear beside content the brand rules out? | Generic or templated content that is harmless but worth little |
| MFA review | Does the site exist mainly to resell attention? | Bots on otherwise good sites, which is why it does not replace IVT filtering |
What does an invalid traffic review look like on a real plan?
Fictional example
A home water treatment brand in Indiana and Iowa
A fictional Indianapolis company that sells water softeners and filtration systems runs a display flight in Indiana and Iowa beside home improvement and household advice content. After the first month the trading desk compares the buying platform's invalid activity adjustments with the verification partner's report.
Most domains show low general and sophisticated invalid rates. One home tips site has a low SIVT rate but a decision rate far below the rest of the buy, because most of its impressions ran in an environment the partner's tag could not read. The desk records that site as undecided rather than clean and asks the seller about the gap.
A second site shows almost no invalid traffic, yet its articles are long slideshows with ads that reload every few seconds, and sponsored link units on other sites point to it. The desk sends that site to the MFA review, not the fraud log, because the problem is the business model rather than the visitors.
Where does PlacementGraph fit next to invalid traffic filtering?
What should a buyer ask a measurement partner about invalid traffic?
Invalid traffic questions for a verification or measurement partner
- The partner states which of its IVT metrics are MRC accredited, and for which environments.
- GIVT and SIVT rates are reported separately, not as one combined figure.
- A decision rate is shown next to every invalid rate, by desktop and web, in-app and OTT.
- The partner explains how SIVT found after a report is handled and disclosed.
- Purchased traffic, where known, is reported as its own segment.
- The report says which IVT definition it follows, MRC or another taxonomy.
Frequently asked questions
Is all invalid traffic fraud?
No. Invalid traffic includes innocent activity such as search engine crawlers, prefetching browsers and monitoring tools, which the MRC lists under general invalid traffic. Fraud describes deliberate schemes, such as hijacked devices or stacked ads, which fall under sophisticated invalid traffic. Both are removed from counts, but only some of it involves intent to deceive.
Do AI crawlers and bots count as invalid traffic?
A crawler is non-human traffic, so impressions it generates should not count as valid. Declared crawlers can be filtered with lists as general invalid traffic. The MRC also encourages reporting bots present with the publisher's permission, such as for measurement or brand safety, as a separate subset. Bots that disguise themselves as people fall under sophisticated invalid traffic.
Does invalid traffic apply to connected TV and apps?
Yes. The MRC standards apply across digital environments and ask for decision rates by desktop and web, in-app and OTT. The April 2024 interim update added requirements on domain and inventory mismatches and bundle ID spoofing in connected TV, where the app declared in a request does not match the app that shows the ad.
What is a normal invalid traffic rate?
There is no published standard rate for buyers to target, and rates depend on the measurer, its decision rate and the environment. Compare a partner's figures over time and across your own placements instead of against a general number, and treat a sudden jump on a few domains as a reason to investigate.
Sources
- Invalid Traffic Detection and Filtration Standards Addendum (IVT 2.0), Update (Final) (opens in a new tab)Media Rating Council, June 2020
- 2024 IVT Interim Updates (opens in a new tab)Media Rating Council, April 24, 2024
- Standards and Guidelines (opens in a new tab)Media Rating Council
- Managing invalid traffic (opens in a new tab)Google Ads Help
- Leading Trade Groups Define “Made For Advertising” Websites (opens in a new tab)ANA (Association of National Advertisers), September 26, 2023
- Certifications (opens in a new tab)TAG (Trustworthy Accountability Group)
Researched and written by PlacementGraph Research. Read how we research, source and update these pages. Corrections: diana@placementgraph.com.