TapCub is live — analytics, insights and live chat, free on one platform
Glossary

What is ad fraud? Types and how traffic data reveals it

Ad fraud is the deliberate creation of fake ad clicks, impressions or installs. The types, how it shows in your traffic data, and a worked cost example.

Xu An Reviewed by Lucas Hayes Also called advertising fraud Last updated Nov 21, 2026 Glossary

Quick answer

Ad fraud is the deliberate creation of fake ad activity, such as clicks, impressions or installs, to collect money from advertisers. It is the intentional part of invalid traffic. In your analytics it shows as clicks that never arrive, near-zero engagement and no conversions from one source.

Formula: Invalid traffic rate = invalid clicks ÷ total clicks

What is ad fraud?

Ad fraud is the deliberate creation of fake ad activity, such as clicks, impressions, installs or conversions, to collect money from advertisers or to drain a competitor's budget. The wider category is invalid traffic: Google defines it as traffic that does not represent genuine interest in your business, and it includes accidents such as clicks caused by ads placed too close to navigation buttons. Fraud is the deliberate part of that category; the accidental part costs you money too, but nobody planned it.

For a site owner, the practical definition is simple: you paid for a visit, and no person who could ever buy came. Advertisers pay for clicks and impressions, so a click made by a script costs the same as one made by a customer. That is the whole incentive. TapCub does not buy, bid on or place ads, so this page is about what you can see after the click, in your own traffic data. Two related terms are the bot farm (a place that generates fake engagement at scale) and the botnet (a network of compromised devices run by an attacker).

Types of ad fraud and invalid traffic

The Media Rating Council splits invalid traffic into two tiers, and Google's help pages use the same split. General invalid traffic (GIVT) is caught by lists and routine checks; sophisticated invalid traffic (SIVT) needs deeper analysis.

TypeTierWhat it isWhat it tends to look like in your data
Data-center trafficGIVTRequests from servers whose IPs are tied to invalid activitySessions from hosting providers, no real browsing
Known crawlersGIVTPrograms that declare themselves non-humanNamed bots in the user agent; counted separately by good analytics
Irregular patternsGIVTAuto-refresh, duplicate clicksMany near-identical hits in seconds
Automated browsingSIVTScripts that do not declare themselves, such as botnet trafficSessions that look human at first glance; zero conversions at scale
False representationSIVTAds shown on a different site, device or country than claimedTraffic from a country you did not target
Manipulated behaviourSIVTPrograms that trigger clicks or installs without consentClicks with no matching visit
Incentivised behaviourSIVTPeople paid to clickReal people, no interest, no purchases

Google's Campaign Manager 360 help page lists these three GIVT and six SIVT categories (the table above merges the last two SIVT categories, misleading interface and undisclosed classification, into the rows that look most like them in your data). Other well-known forms, such as clickjacking (invisible ads that trick users into clicking) and ad stacking (several ads layered so only the top one is visible), appear in Google Ads' list of fraudulent placements.

How traffic data reveals it

Fraud is hard to prove from one number, but it leaves a pattern across several. Sample data: a campaign spent 5,000 on 10,000 clicks at 0.50 each, split over three traffic sources.

SourceClicksSessions seen in analyticsArrival rateConversionsSpend
Search brand terms3,0002,90096.7%901,500
Social campaign4,0003,10077.5%302,000
Partner network X3,00080026.7%01,500
Total10,0006,80068.0%1205,000

Four signals stand out for network X:

  1. Arrival rate: clicks that never become sessions. Some gap is normal (people close the tab, scripts are blocked), but 26.7% against 77–97% elsewhere is not.
  2. Behaviour: almost no engagement, very short visits and a bounce rate near 100% on the sessions that do arrive.
  3. Zero conversions at a volume where the other sources convert.
  4. Odd concentrations: most sessions from one country, device model, hour of the night or hosting network.

Calculate the damage: reported cost per conversion is 5,000 ÷ 120 = 41.67. Without network X it is 3,500 ÷ 120 = 29.17, so the suspicious source inflates the cost per conversion by 41.67 ÷ 29.17 − 1 = 42.9%, and 1,500 of 5,000 (30.0%) of the spend went to a source that produced nothing.

A single signal can be innocent; low arrival can mean a slow page, and zero conversions can mean a bad audience rather than a bot. Treat three or more of the four signals together as a reason to ask the ad platform for an investigation, not as proof.

Formula and the cost of invalid traffic

Invalid traffic rate = invalid clicks ÷ total clicks. If 1,000 of the 10,000 clicks were invalid, the rate is 10.0%, and at 0.50 per click that is 500 of wasted spend. Google says invalid traffic caught before billing is removed from your costs, and invalid activity found after billing is credited in the next billing cycle where possible. Its investigation requests cover activity from the last 60 days, so check campaigns monthly, not yearly.

Common misunderstandings

  • "A high bounce rate means fraud." Not by itself: a fast page with a clear answer also has high bounce. Compare with the same source on other days.
  • "Bots are always caught by analytics." Declared crawlers are easy to separate. Bots that pretend to be ordinary browsers can still look like visitors, which is why source-level patterns matter.
  • "The ad platform's number is the true number." Clicks are counted by the platform and sessions by your analytics; the two will never match exactly, so track the gap rather than demand zero.
  • "Fraud only happens on cheap networks." It happens wherever there is money per click or per install.

What to send the ad platform

If the signals add up, send the platform a short evidence pack rather than a complaint: the date range (within the last 60 days), the campaign and source, platform clicks against analytics sessions, the arrival rate, engagement figures and the spend involved. In the sample, that is "network X, 3,000 clicks, 800 sessions (26.7%), no conversions, 1,500 spent". Ask for an invalid-activity review and keep your own table, because the platform's decision may take several business days and the numbers you recorded are the only copy of what you saw.

How to do it in TapCub

Run the network X example in TapCub:

  1. Tag every ad link with source and campaign parameters, then open Web analytics › Sources to see sessions by source and campaign next to your click counts from the ad platform.
  2. Compare the arrival rate by dividing sessions by platform clicks for each source.
  3. Check behaviour. Look at bounce rate, average duration and goals by source in Web analytics › Goals.
  4. See recognised bots. Web analytics › Bots lists crawlers TapCub identified, their operator, purpose and request count; they are kept out of visitor and conversion numbers. Bots posing as ordinary browsers are not identified this way, so the source-level signals matter.
  5. Import the spend. In All-platform analytics › Business › Channel ROI import cost by day, channel, source and campaign to see cost per click and cost per acquisition by source.

TapCub does not buy ads, block clicks or request refunds: it shows the pattern so that you can take it to the ad platform.

Sources

Was this explanation clear?

We use your answers to fix and expand this page.

Xu An

Customer Experience Consultant

Writes about live chat, visitor conversations, and self-service, sharing practical ways to make customer communication clearer and more effective.

FAQ

Questions people ask

Short answers to the follow-up questions that come up most.

Still have a question?

Chat with the team behind TapCub — we usually reply within a few hours on business days.

What is ad fraud in simple terms?

Ad fraud is when someone, usually a script or a paid network, creates fake clicks, views or installs on an ad so that money is paid for activity no interested person produced. The advertiser pays the same as for a real click.

What is the difference between ad fraud and invalid traffic?

Invalid traffic is the wider group: any traffic that does not represent genuine interest, including accidental clicks. Ad fraud is the deliberate part. Google and the Media Rating Council split it into general (GIVT) and sophisticated (SIVT) invalid traffic.

How can you tell if you have ad fraud?

Compare platform clicks with analytics sessions by source, then look for low arrival rates, near-zero engagement, odd concentrations by country or hour, and zero conversions. In the sample, one network had a 26.7% arrival rate and no conversions.

Can you get a refund for ad fraud?

Sometimes. Google says invalid traffic found before billing is removed from costs and invalid activity found after billing is credited in the next cycle where possible; investigation requests cover the last 60 days.

See this number on your own site

Add one script tag for web and apps, and the first reports fill in within minutes.

See your data clearly. Find your growth.

Every click, backed by data. Install one line of code and see your first numbers in a minute.

No credit card · Free plans for analytics and chat · Cookieless analytics