Invalid Traffic (IVT) and Ad Click Verification

Invalid traffic (IVT) is the umbrella term for ad impressions and clicks that do not reflect genuine interest but can still end up in a report, an invoice or a publisher's earnings. For an advertiser it means budget that never reached a real customer. For a publisher it can mean corrected earnings or a restricted ad account. Three questions follow from that. How do I know a click was valid, what does bot traffic look like in an ad report, and what evidence should I ask a seller for?

This guide takes its definition of invalid traffic from Google's AdSense and Google Ads help pages, and its split between GIVT and SIVT from the Media Rating Council's (MRC) June 2020 invalid traffic standards addendum. It then walks through practical ad click verification, record keeping and the questions worth adding to a contract. No tool, service or platform is promised to give full protection, and the numerical values in this article are quoted only from the MRC addendum. Anything not attributed to a source is the author's judgement, not a quoted rule.

What Is Invalid Traffic (IVT)?

How Google defines it

Google's AdSense invalid traffic page (Opens in a new tab) describes invalid traffic as any click or impression that may artificially inflate an advertiser's costs or a publisher's earnings. The page says the term covers intentionally fraudulent traffic as well as accidental clicks. So invalid traffic is not only a malicious bot attack. Unintended taps caused by a badly designed placement fall under the same heading.

The examples on that page include publishers clicking their own live ads, repeated clicks or impressions from one or more users, publishers encouraging clicks (including wording that urges people to click ads), implementations that may cause a high volume of accidental clicks, and automated clicking tools, robots or other deceptive software. The list is a useful frame for the rest of this article because it shows what both buyers and publishers need to question.

The advertiser's view: invalid clicks

The Google Ads invalid clicks page (Opens in a new tab) tells the same story from the buyer's side. It defines an invalid click as a click that does not represent genuine interest in your business. Its examples are accidental clicks caused by poor ad placement, automated traffic from bots or deceptive software, and manual clicks that raise ad costs or inflate publisher profits. The page also says interactions found to be invalid are filtered out of campaign metrics and removed from billing, that you are not charged for invalid clicks or impressions, and that invalid clicks found after billing are credited on the next invoice where appropriate and possible.

That description applies to Google Ads' own inventory. If you buy ad space directly from a publisher, do not assume the same filtering and credit exist by default. Which protections apply, under what conditions and measured by whom needs to be written down with the seller.

Who is responsible?

The AdSense page accepts that a third party can generate invalid traffic without the publisher's knowledge, and still states that ultimately it is the publisher's responsibility to make sure traffic on their ads is valid. According to the page, an account with high levels of invalid traffic may be suspended or closed, and if Google cannot verify traffic quality it may limit or disable ad serving. Invalid clicks can also cause a difference between estimated and finalized earnings. So this is a revenue question for publishers, not only a budget question for advertisers.

GIVT and SIVT: The Two Categories in MRC's Standards

General invalid traffic (GIVT)

The MRC's June 2020 invalid traffic addendum (Opens in a new tab) sets up two categories. The first is General Invalid Traffic (GIVT): traffic identified through routine filtration, either by applying lists or by standardized parameter checks. The document's key examples include known invalid data-center traffic, bots and spiders, non-browser user-agent headers, pre-fetch traffic where the ad was never subsequently accessed by a valid user, invalid ad sizes such as 0x0 and 1x1, and sessions that cannot render images.

The addendum says known GIVT must be removed from monetized counts and metrics. Every accredited or certified digital measurement organization must apply GIVT detection. GIVT techniques are not separately accredited, while SIVT detection can be independently accredited or certified.

Sophisticated invalid traffic (SIVT)

The second category is Sophisticated Invalid Traffic (SIVT). The addendum describes it as harder-to-detect situations that need advanced analytics, multi-point corroboration or significant human intervention. Its examples include automated browsing from dedicated or hijacked devices, incentivized human activity such as click farms, manipulated activity such as forced new tabs and tricking users into accidental clicks, falsified measurement events such as fake clicks or impressions, domain and app misrepresentation, bots masquerading as legitimate users, and hidden, stacked or transparent ad serving.

For a buyer this becomes a concrete question. If a vendor's report rests only on list-based filtering, sophisticated behaviour that is not on any list could be sitting inside the numbers. The addendum addresses exactly that: where the measurement organization of record applies only GIVT, buyers are encouraged to consider adding SIVT capability through a third party or an alternate method. That is a recommendation in the document, not a requirement.

The IAB Tech Lab Spiders and Bots list

The MRC addendum stresses uniform lists for GIVT processes and names the IAB/ABC International Spiders and Bots list as an example. According to the IAB Tech Lab list page (Opens in a new tab), the list is two text files, one of valid browsers and user agents and one of known robots, used together in a dual-pass filtering process. The page says the list is updated monthly, sold as a 12-month subscription and does not require membership (the page itself was last updated on 24 April 2025). It also says the list helps companies meet the MRC's GIVT detection requirement. A seller saying it uses such a list does not prove its whole filtering process is sufficient, but asking which bot list it relies on is a concrete, fair question.

Ad Click Verification: What Should a Buyer Check?

What verification is for

Ad click verification is the process of checking that a click was a deliberate human action, that it landed on the right page and that it matches the number you are billed for. There is rarely one definitive method. It is usually a comparison of several sources: the seller's report, your own web analytics, your server logs and, where available, an independent measurement provider. None of them is conclusive alone. The aim is to spot a mismatch early and ask about it.

A step-by-step verification routine

The routine below is a working pattern for buying a banner or sponsored placement. The sequence is this article's suggestion, not an MRC or Google requirement.

1. Before the campaign starts, put the measurement definition in writing: which event counts as a click, which tool measures it and when an impression is counted.
2. Make the landing address trackable. Give every seller or placement its own campaign tag (a UTM parameter) so incoming sessions can be separated in your own analytics.
3. In the first days, compare the seller's report with your analytics. The MRC addendum likewise encourages buyers and sellers to talk about early performance, suggesting the first part of the campaign.
4. Record the gap between clicks and sessions and ask what explains it. A gap does not always point to invalid traffic; the next section lists legitimate causes.
5. Look at session quality: time on page, scrolling, a second page view and conversion events. These are flags that call for review, not proof.
6. Log anything suspicious with date, time, source, placement and a screenshot, and send it to the seller in writing.
7. At the end, ask the seller which filtering the final report went through, and how much activity was removed and by what method.

How to read the headline metrics is covered in measuring banner ad performance. This article is about asking whether those numbers rest on reliable data in the first place.

A fictional working routine

The routine below is fictional. It does not describe a real campaign, a measurement result or the behaviour of any party. Imagine an e-commerce brand buying a two-week banner placement on a news site. Before the campaign starts, the brand agrees in writing with the seller which event counts as a click and which tool measures it. It adds its own tag to the landing address and asks for a daily report. In the first days it compares the report with its analytics. If it sees a gap, it does not accuse; it asks the seller for the counting definition and the filtering applied, and checks its own analytics setup. If the gap stays unexplained, it sends the finding in writing with date, source and a screenshot. What matters in this routine is not an outcome but that the process runs early and leaves a record.

Which records are worth keeping?

When a dispute arises, dated records carry far more weight than a verbal objection. The table summarizes what you can consider keeping during a campaign.

Server logs can contain IP addresses and other personal data. When you keep them, follow your own privacy policy and applicable law on retention periods and access.

Why Clicks and Sessions Rarely Match

Not every gap is invalid traffic

The click count in a seller's report and the session count in your analytics will usually differ. There are technical reasons that have nothing to do with bad faith. A user may click and leave before the page loads, the analytics script may be blocked or fail to run, a consent choice may stop the session from being recorded, redirects may strip parameters, or the two systems may count the click at different moments. These causes come from field experience and are not tied to any rate or threshold. What matters is to eliminate causes one by one instead of attributing the gap to a single explanation.

So the first reaction should be a question, not an accusation. Review the seller's counting definition together with your own analytics setup. Page load speed matters here too: a slow landing page can stop valid clicks from being recorded as sessions.

Signals that justify a closer look

The signals below do not prove invalid traffic on their own. They overlap with the examples named by MRC and Google, which makes them worth investigating.

• Repeated clicks from the same source in a short period (Google lists repeated clicks among its examples).
• Placements next to navigation or in spots where a thumb can easily tap by mistake, and designs that invite accidental clicks.
• Clicks with no page interaction at all, or sessions that all behave identically.
• Unusual user-agent distributions or non-browser headers (an MRC GIVT example).
• Records showing the ad appeared on an unexpected domain or app (the MRC lists domain and app misrepresentation among SIVT examples).
• Evidence that the ad was served hidden, stacked or transparent (an MRC SIVT example).

If a review leaves you unsure, do not settle it by guesswork. Ask an independent measurement provider, or ask for the seller's own audit record. More general audience and traffic checks on the buyer side are in the guide to verifying a publisher's audience.

For Publishers: Preventing Invalid Traffic

Your own ads and click encouragement

The AdSense page lists a publisher clicking their own live ads, and wording that urges users to click ads, among its examples of invalid traffic. A simple team rule helps: nobody clicks live ads, even for testing, and page copy never asks readers to click an ad. Labelling an ad and urging a click are different things. A label informs the reader, a call to action steers them.

Placement and accidental clicks

Both Google pages mention accidental clicks, and the Google Ads page says they can come from poor ad placement. The practical consequence for a publisher is to keep ads away from navigation elements, areas people tap by mistake while scrolling and spots where content may shift under a finger. The article on ad viewability and IAB standards complements this for the visibility and measurement side. Viewability and invalid traffic are different concepts, and the MRC addendum notes that its non-rendering category is agnostic of ad visibility.

Caution with third-party sources

Google acknowledges that third parties can generate traffic without the publisher's knowledge. That makes it necessary to know where your traffic comes from. If you pay for traffic, get in writing how the source works, which clicks are human actions and how that is measured. A sudden unexplained spike deserves investigation before ad revenue ever drops. Since the AdSense text puts ultimate responsibility on the publisher, this check is a duty and not a preference.

Summarizing your audience and traffic sources on one page also makes a buyer's job easier. An advertiser can then see your basic quality signals before sending detailed questions, and the publisher quality checklist is a helpful starting point for deciding what to show.

Questions for the Seller and Contract Clauses

What to ask before you buy

The questions a buyer should ask can be derived from the points the sources raise. The list below is a checklist. Having an answer to each item helps a buyer make an informed decision, but nobody guarantees the answers are accurate, so verification still means testing what the seller says.

• Do you apply invalid traffic filtering? If so, how do you handle GIVT and SIVT separately?
• Is the filtering done by you or by an independent measurement provider? What is the provider's name and method?
• Are the clicks and impressions you report before or after filtering? Is removed activity shown separately in the report?
• Which bot lists or classification methods do you use?
• On which pages and positions does the ad appear? Is any position prone to accidental clicks?
• Does your traffic include purchased or incentivized sources?
• What process do you follow when a click or impression discrepancy comes up?

None of these questions implies a particular answer is true. Still, any question left unanswered or vague is a legitimate reason to postpone the purchase. The full buying process is laid out step by step in the guide to buying ad space on websites.

Which clauses belong in the agreement?

Verbal answers are not a contract, so they need to be recorded. Consider including: which tool and definition will be used for measurement, how activity is judged invalid, the time allowed for objections, sharing of records with the buyer, the remedy for confirmed invalid activity (correction or refund) and the process for resolving a dispute. The legal validity and wording of each clause depend on the parties drafting the agreement. For a more detailed list see the ad space contract checklist, and take legal advice where needed.

MRC's framework for discrepancies

The MRC addendum also sketches a framework for measurement disagreements. For that process it treats a significant or material discrepancy as generally greater than five percent of the affected metric, and it adds that relative thresholds may be combined with absolute dollar values. That figure is how the MRC defines its own discrepancy resolution process. It does not mean that any click or impression gap is acceptable across the industry, and your contract is not obliged to use the same threshold.

The document's practical advice comes earlier in the process. Buyers and sellers are encouraged to share information about their measurement organizations before a campaign starts, to discuss initial performance early (the suggested window is the first twenty percent of the campaign) and to use consistent language about invalid traffic, with the TAG taxonomy suggested for that purpose. That way a problem is caught while it is still easy to correct.

Fake Traffic, Viewability and Performance: Three Concepts People Mix Up

They are not the same thing

These three ideas are often confused. Invalid traffic, in the ad context, is activity that does not represent genuine interest in a click or impression. Fake traffic more often means visitors manufactured to inflate a site's numbers or mislead authority metrics, a subject covered separately in fake traffic and inflated SEO metrics. Viewability measures whether an ad could actually be seen on screen, and it is independent of whether a click was valid.

The distinction matters in practice. One publisher can have real, high traffic and a placement nobody sees. Another can have a visible ad whose clicks come from accidental taps. Each needs its own check, and a good result on one says nothing about the other.

Which metric answers which question?

When you read the table, mind the order. First decide which concept answers your question, then look at the data source that fits it. A good viewability report does not show that the clicks were valid, and a clean click report does not show that the ad was seen long enough. Resting a purchase on one indicator is weaker than collecting an answer to each question from its own source.

Common Mistakes

On the buyer's side

The first mistake is looking only at the click count. A high number may not reflect genuine interest, and by Google's definition a click that does not represent genuine interest is invalid. The second is never asking who does the measuring, since the seller's own report is not enough evidence by itself. The third is noticing the problem only after the campaign ends, when early comparison would have made a correction easier. The fourth is blaming invalid traffic for every gap; accusing before ruling out technical causes damages the relationship and the outcome.

On the publisher's side

A frequent publisher error is welcoming a traffic spike without asking where it came from. Another is putting an ad in a spot prone to accidental taps and reading the extra revenue as performance; because Google Ads treats accidental clicks from poor placement as invalid, that revenue can be corrected later. A third is letting staff click the site's own ads, which AdSense names explicitly as an invalid traffic example.

Frequently Asked Questions

What is invalid traffic (IVT)?

Invalid traffic is any click or impression that may artificially inflate an advertiser's costs or a publisher's earnings. Google's definition covers intentional fraud as well as accidental clicks. Bots, automated tools, repeated clicks, a publisher clicking their own ads and wording that encourages clicks are among the examples.

What is the difference between GIVT and SIVT?

According to the MRC, GIVT is traffic identified through routine filtration with lists and standardized parameter checks; bots, spiders and known data-center traffic are examples. SIVT covers harder situations that need advanced analytics, multi-point corroboration and human intervention; hijacked devices and falsified measurement events are examples.

Can I get a refund for invalid clicks?

For Google Ads, the help page says you are not charged for invalid clicks and impressions, and that those found after billing are credited on the next invoice where appropriate and possible. That description applies to Google Ads. For ad space bought directly, refund and correction terms need to be written into your agreement.

How can I tell bot traffic in an ad report?

There is no single definitive sign. Repeated clicks, sessions with no interaction, unusual user-agent distributions or unexpected source domains are signals that call for review. They are not proof, and they should be compared with the seller's report, your own analytics and, if needed, independent measurement.

Who performs ad click verification?

A buyer can run a first check with their own analytics and tagged campaign addresses. For deeper detection, the seller's filtering or an independent measurement provider can step in. Where a measurement organization applies only GIVT, the MRC addendum encourages buyers to consider adding SIVT capability through a third party or an alternate method.

What can a publisher lose because of invalid traffic?

According to the AdSense page, invalid clicks can cause a difference between estimated and finalized earnings, and an account with high levels of invalid traffic may be suspended or closed. If Google cannot verify traffic quality it may limit or disable ad serving.

Conclusion and Next Step

Invalid traffic is not just a bot attack. It is any click or impression that does not represent genuine interest. Google keeps intentional fraud and accidental clicks under one roof, while the MRC separates GIVT from SIVT by how hard the traffic is to detect. For a buyer the real work is to write down the measurement definition before the campaign, set up tagged tracking and ask the seller how filtering works. For a publisher it is to avoid clicking your own ads, never encourage clicks and know your traffic sources. No source says these steps remove invalid traffic entirely; the goal is to make the risk visible and manageable.

This article does not claim that Akua Adforce filters invalid traffic or offers any protection. When you browse the ad space listings on Akua, put the questions above to the relevant seller; each listing's scope and terms are written on the related product and policy pages. Ask for measurement and filtering details in writing, and keep the answers with your campaign records.