Executive summary
- 1The most-cited cost of invalid traffic is three years old: according to a 2023 estimate from Juniper Research, 22% of global digital ad spend, or $84.2 billion, was lost to ad fraud in 2023 [3], while the ANA's Q2 2025 benchmark puts programmatic waste at $26.8 billion a year [5].
- 2Bots now exceed 53% of all web traffic according to Imperva's Bad Bot Report 2026 [1], humans generated only 47% of HTML requests on Cloudflare-protected sites by December 2025 [6], and Fastly measured automated traffic at 37% of activity on its network in mid-2025 [7].
- 3Protection is the variable that matters most: TAG held invalid traffic under 1% in US certified channels for a fourth straight year [8], while unprotected European video and display channels ran at 10.36% and carried more than 10 times the relative loss [2].
- 4The platforms' own reports show the scale of the supply: Google removed 8.3 billion ads in 2025, 1.29 billion of them for abusing the ad network [4]; Meta actioned 1.1 billion fake accounts in Q4 2025 and estimates 5% of daily active people are fake [9]; LinkedIn stopped 88.9 million fake accounts at registration in six months [10].
- 5Beyond the click, AppsFlyer finds organic now carries 52% of fraudulent installs and affiliates run a 36x higher fraud rate than self-reporting networks [11].
- 6India's digital ad spend reached Rs 71,621 crore in 2025, 59% of the market [12]; India accounts for nearly 28% of global Android install fraud [11]; and ClickTrust audits find 14 to 42 percent of paid traffic invalid depending on channel.
What is invalid traffic and how is it measured?
Invalid traffic is any ad click, impression or interaction that does not come from a real person with real interest in the ad. The Media Rating Council (MRC), which accredits ad measurement, defines it 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" in its June 2020 standards update [13]. Google's Ad Traffic Quality site puts it more plainly: "any interaction that doesn't come from real people with real interest in an ad" [14].
The MRC has split invalid traffic into two tiers since its October 2015 Invalid Traffic Detection and Filtration Guidelines Addendum [15]. General Invalid Traffic (GIVT) "includes traffic identified through routine and list-based means of filtration", such as bots, spiders and crawlers, non-browser user agents, and pre-fetch or pre-rendered traffic [15]. Sophisticated Invalid Traffic (SIVT) "includes traffic identified through advanced analytics, multipoint corroboration, human intervention", such as hijacked devices, ad tags or creative, adware, malware and misappropriated content [15]. The 2020 update added more SIVT categories, a required "decision rate" metric and formal machine-learning and human-review requirements [13].
Why the GIVT and SIVT split decides your bill
GIVT is cheap to catch and is mostly filtered before you pay. SIVT is what reaches your budget, because it needs behavioral analysis and corroboration across signals rather than a list. Google draws the same line in practice: some invalid traffic is accidental, such as a badly placed ad, and some is deliberate, such as botnets or competitors clicking to waste budgets [14].
How much paid media is lost to invalid traffic?
The honest answer is a range, and the most-cited number is three years old. According to a 2023 estimate in Juniper Research's whitepaper Quantifying the Cost of Ad Fraud 2023-2028, 22% of global digital ad spend, or $84.2 billion, was lost to ad fraud in 2023, with losses forecast to reach $172.3 billion by 2028 [3]. The same study estimated that 17% of desktop clickthroughs were illegitimate, with fraudulent clickthroughs growing from 37 billion in 2023 to more than 65 billion by 2028 [3].
The World Federation of Advertisers' long-standing projection, as quoted by PPC Land in March 2025, is that ad fraud is likely to exceed $50 billion globally in a year [16]. That projection originates in 2016 and no newer WFA figure could be verified, so treat it as a floor rather than a current measurement.
The freshest allowed numbers are narrower and come from the buy side. The ANA's Q2 2025 Programmatic Transparency Benchmark found $26.8 billion of global media value is still lost each year to programmatic inefficiencies [5]. In Europe, TAG's 2025 European Ad Fraud Savings Report estimated that anti-fraud standards saved €3.45 billion in 2023, a 69% reduction, on a €48.3 billion digital ad market [2]. In the US, TAG put the 2023 saving from industry anti-fraud programs at $10.8 billion [8].
- $84.2B
- lost to ad fraud in 2023, per Juniper Research's 2023 estimate, 22% of global digital ad spend[3]
- $26.8B
- of global media value still lost each year to programmatic inefficiencies, ANA Q2 2025[5]
- €3.45B
- saved by anti-fraud standards in Europe in 2023, a 69% reduction in losses[2]
| Publisher | Data period | What it measures | Headline figure |
|---|---|---|---|
| Juniper Research [3] | 2023, forecast to 2028 | Modeled share of global digital ad spend lost to fraud | 22% of spend, $84.2B in 2023, $172.3B by 2028 |
| WFA via PPC Land [16] | 2016 projection, quoted 2025 | Projected global ad-fraud losses | Likely to exceed $50B a year |
| ANA [5] | Q2 2025 | Programmatic spend of benchmark participants | $26.8B lost annually to inefficiencies |
| TAG, United States [8] | Jan to Jun 2024 | 745B impressions in TAG Certified Channels | IVT under 1% for a fourth year; $10.8B saved in 2023 |
| TAG, Europe [2] | 2023 | €48.3B market, protected vs unprotected channels | 10.36% unprotected vs about 1% protected; €3.45B saved |
| Juniper Research [17] | 2025, forecast to 2030 | Scam-ad impressions on European social media | Nearly 1 trillion, about 10% of social advertising |
Population and timing explain the range. Juniper models all digital spend, the ANA measures the programmatic spend of its member advertisers, and TAG measures impressions inside channels that already filter, so a certified-channel rate is what survives filtration while a modeled rate is what arrives at the door. ClickTrust audits of individual accounts typically find 14 to 42 percent of paid traffic is invalid depending on channel, above these averages because audits start with accounts that already show a symptom.
Social platforms have their own waste line. Juniper Research estimated in February 2026 that nearly 1 trillion scam-ad impressions reached European social media users in 2025, around 10% of all social media advertising [17]. Scam ads are not invalid traffic by MRC definition, but they draw on the same fake accounts and rented devices.
Protected vs unprotected channels: what certification changes
Invalid traffic inside certified channels runs near 1%; outside them it runs above 10%. TAG's October 2024 analysis of 745 billion impressions found invalid traffic held under 1% in TAG Certified Channels in the United States for the fourth consecutive year [8]. A TAG Certified Channel is one where the agency, buying platform and selling platform in a transaction are all Certified Against Fraud [2].
Europe shows the gap. TAG's 2025 European Ad Fraud Savings Report estimated a 10.36% invalid-traffic rate in unfiltered video and display channels, against about 1% in protected channels [2]. About 76% of European 2023 spend, €36.8 billion, flowed through protected channels and lost roughly €368 million; the 24% that did not, €11.5 billion, likely sustained €1.19 billion of losses, more than 10 times the relative level [2]. TAG estimates a further €1.075 billion a year could be saved if that spend moved into protected channels [2].
Source: TAG, 2025 European Ad Fraud Savings Report (2025, 2023 data) [2]
| Channel type | Share of spend | Spend | Invalid traffic rate | Estimated loss |
|---|---|---|---|---|
| Protected (TAG certified) [2] | 76% | €36.8B | About 1% | €368M |
| Unprotected video and display [2] | 24% | €11.5B | 10.36% | €1.19B |
Quality governance moves performance, not only waste
The ANA's Q4 2025 Programmatic Transparency Benchmark found advertisers enforcing quality governance converted 56.7% of programmatic spend into benchmark-qualified impressions against 37.5% for lower performers, with case studies showing nearly 40% reductions in cost per conversion [19]. The share of advertisers above the 50% threshold fell from 50.0% in Q3 to 43.8% in Q4, so the gains are not locked in [19]. Private marketplaces carried more than 92% of median spend and CTV about 40% of programmatic spend [19].
Made-for-advertising sites are shrinking but not gone
MFA inventory is not invalid traffic by MRC definition, but it is spend that reaches no one worth reaching. The ANA's 2023 Programmatic Media Transparency Study found MFA sites made up 21% of impressions and 15% of ad spend, and that the average campaign ran on 44,000 domains [20]. By Q2 2025 the ANA's median MFA spend had fallen to 0.8%, down from 15% two years earlier, yet top-quartile marketers still spent up to 28.7% on MFA domains [5]. The programmatic supply notes cover placement-level controls.
What do the platforms themselves report?
The platforms publish more about invalid activity than most advertisers read. Google's 2025 Ads Safety Report says it blocked or removed more than 8.3 billion ads, restricted 4.8 billion more and suspended 24.9 million advertiser accounts in 2025, stopping more than 99% of policy-violating ads before they served [4]. The single largest enforcement category was Abusing the Ad Network at more than 1.29 billion ads [4]. Scam-related policies accounted for 602 million ads and 4 million suspended accounts, and Google actioned more than 480 million publisher pages [4].
Meta reports from the account side. In its March 2026 Code of Practice filing to the EU, Meta reported acting against 692 million fake accounts on Facebook in Q3 2025 and 1.1 billion in Q4 2025, and estimated fake accounts at about 4% of worldwide daily active people in Q3 and 5% in Q4 [9]. Its September 2025 filing reported 1 billion fake accounts actioned in Q1 2025 and 687 million in Q2, at roughly 3% and 4% of monthly active users [21]. Meta's H1 2026 Integrity Reports add that it removed more than 159 million scam ads in 2025, 92% before anyone reported them [22].
LinkedIn's Community Report for July to December 2025 shows 88.9 million fake accounts stopped at registration, 24.4 million restricted proactively before any member report and 361,300 restricted after reports, with 99.7% of fake accounts stopped before a member report [10]. LinkedIn also removed 68.3 million spam or scam items proactively in the same six months [10].
| Platform | Period | What it reports | Figure |
|---|---|---|---|
| Google, global [4] | 2025 | Ads blocked or removed; advertiser accounts suspended | 8.3B+ ads; 24.9M+ accounts; 1.29B+ ads for abusing the ad network |
| Google, India [23] | 2025 | Ads blocked or removed; advertiser accounts suspended | 483.7M ads; 1.7M accounts |
| Meta, Facebook [9] | Q4 2025 | Fake accounts actioned; estimated prevalence | 1.1B accounts; about 5% of daily active people |
| Meta [22] | 2025 | Scam ads removed | 159M+, 92% before anyone reported them |
| LinkedIn [10] | Jul to Dec 2025 | Fake accounts stopped at registration | 88.9M; 99.7% of fake accounts stopped proactively |
Source: Meta, Code of Practice on Disinformation reports (September 2025 and March 2026) [21, 9]
Read these as capacity, not residue. A platform that removes 1.1 billion fake accounts in a quarter is describing how large the supply of fake identities is. Fake accounts at 5% of daily active people is the share Meta estimates it has not yet removed, and that share can click, like and submit a form on your ad [9]. None of these reports publishes the invalid share of billed clicks on an advertiser's account, which is the number a CFO needs.
What does Google refund, and what does the Invalid Traffic card show?
We don't charge for invalid traffic.
Invalid traffic Google detects before billing is never charged. Google Ads Help says invalid traffic detected after the billing period ended is returned as credited cost, and that advertisers can request an investigation of activity within the last 60 days through the Click Quality Form, which typically takes several business days [25]. The invalid traffic card shows two numbers: estimated savings from automated protections and manual reviews, and credited cost for past invalid activity [25]. On 12 November 2026 Google announced the Invalid Traffic Overview Card, which presents these as "Automated protection" and "Past invalid activity" [24]. Keerat Sharma, Google's VP of Ads Privacy and Safety, said Google had cut advertiser exposure to invalid leads by an average of 30% so far in 2026 [24].
The Invalid Activity Credit Report, found in Report Editor's template gallery, breaks credits down by campaign and network with columns for credited clicks, credited interactions, credited amount and adjusted cost [26]. The same help page says the Click Quality Form covers "invalid clicks, poor-quality form submissions, clicks outside location targeting, or sudden short-term cost increases", so lead quality is explicitly in scope [26].
What the refunds do not cover
Google's credits are real, but they are defined by Google's own detection and reported only in aggregate. The 60-day window closes before most quarterly reconciliations, and an investigation request needs evidence the advertiser must bring [25]. Google sees the click and, increasingly, the lead form; it does not see your app install, your affiliate postback or the bot that returns a week later as direct traffic.
Microsoft's traffic-quality policy for Microsoft Invest, its programmatic buying platform, is unusually candid about the limits of platform-side filtering. It subscribes to the IAB bots and spiders list, de-duplicates clicks per impression, and states that "We block some cloud providers but not all" [27]. It also needs at least 10,000 impressions on a slice of inventory before a click-through or conversion anomaly is treated as significant, so small campaigns sit below the platform's radar [27].
Where else does invalid traffic hit: installs, affiliates, CTV and leads
App installs
AppsFlyer's State of Fraud for Marketers 2026, built on more than 100 billion installs across 246,000 apps from Q1 2025 to Q1 2026, found organic now accounts for 52% of all fraudulent installs, the single largest fraud channel [11]. iOS install fraud fell 33% from 17.5% in Q1 2025 to 11.7% in Q1 2026, dropping below Android's 14 to 15% [11]. Fake installs made up 87% of detected Android fraud and 92% of iOS fraud, and in Android gaming emulator fraud fell 32% while physical device farms grew 72% [11].
Source: AppsFlyer, The State of Fraud for Marketers 2026 (April 2026) [11]
Vertical matters more than platform. AppsFlyer's Android fraud rates run at 7% for gaming, 10% for shopping and 31% for finance, and gambling rose from 49% to 59% over the year, peaking at 64% in Q4 2025 [11]. Fraud is also moving into channels advertisers trust: DSP install fraud rose from 5.6% to 8.9% and owned-media fraud from 3.4% to 11% [11]. In-App Protection covers the SDK side of this problem.
Affiliates
Affiliates account for about 40% of fraudulent installs with a fraud rate near 40% every quarter, while self-reporting networks carry 1.5% of fraudulent installs at a rate around 1%, the 36x gap [11]. Shopping affiliate fraud rose from 27% to 36%, and social-media affiliate fraud ran near 50% [11]. Affiliate programs need per-partner scoring, which is why Lite-MMP Attribution scores each affiliate separately.
- 52%
- of fraudulent mobile installs now arrive through the organic channel[11]
- 36x
- higher install-fraud rate on affiliates than on self-reporting networks, Q1 2026[11]
- 59%
- Android install fraud rate in gambling in Q1 2026, after a 64% peak in Q4 2025[11]
Connected TV
No allowed source in our evidence base publishes a CTV invalid-traffic rate, so this report does not quote one. What is measurable is exposure: CTV was about 40% of programmatic spend among ANA benchmark participants in Q4 2025 [19], and Juniper's 2023 estimate expected in-app advertising to account for 52% of global fraud losses by 2028 [3]. Treat CTV as unprotected programmatic until a certified channel proves otherwise.
Leads and forms
Google's inclusion of poor-quality form submissions in the Click Quality Form scope [26] and its reported 30% reduction in invalid-lead exposure during 2026 [24] are the clearest platform acknowledgements that invalid traffic now reaches the form, not only the click. Meta's estimate that 5% of daily active people are fake accounts [9] and LinkedIn's 88.9 million blocked registrations in six months [10] describe the supply behind fake leads. A form submission from a fake account costs the same as one from a buyer, and sales pays for it a second time.
How does India compare on invalid traffic?
India's exposure is large and growing faster than most markets, but no allowed source in this report publishes an India invalid-click rate, so the comparison rests on spend, enforcement and install fraud. According to the dentsu-e4m Digital Advertising Report 2026 as reported by Social Samosa, India's digital ad spend reached Rs 71,621 crore in 2025, up 19%, or 59% of a Rs 1,21,339 crore total market, with digital projected to take a 70% share by 2027 [12]. Programmatic was the largest line at Rs 30,081 crore, 42% of digital, followed by social media at Rs 21,057 crore, online video at Rs 20,004 crore, e-retail media at Rs 17,601 crore and paid search at Rs 16,581 crore [12].
Other estimates are larger. The FICCI-EY report put digital advertising at INR 947 billion in 2025, 63% of total ad revenue [28], and WPP Media's This Year Next Year forecast, as reported by Social Samosa, expects digital at 68.1% of Rs 2,01,891 crore in 2026 [29].
| Publisher | Measure | Figure |
|---|---|---|
| dentsu-e4m via Social Samosa [12] | Digital ad spend, 2025 | Rs 71,621 crore, up 19%, 59% of total |
| FICCI-EY [28] | Digital ad revenue, 2025 | INR 947 billion, up 26%, 63% of total |
| WPP Media via Social Samosa [29] | Total ad revenue, 2026 forecast | Rs 2,01,891 crore, digital 68.1% |
| Google India [23] | Ads blocked or removed in India, 2025 | 483.7M ads; 1.7M advertiser accounts suspended |
| AppsFlyer [11] | Share of global Android install fraud | Nearly 28%; affiliate share of India's fraud 57% to 65% |
| MHA / I4C [30] | Financial-fraud complaints, 2025 | 24,02,579 complaints; Rs 22,495 crore reported lost |
Enforcement scales with spend. Google's India edition of its 2025 Ads Safety Report says it blocked or removed 483.7 million ads and suspended 1.7 million advertiser accounts in India in 2025 [23], against 8.3 billion ads and 24.9 million accounts globally [4].
The install side is the sharpest signal. AppsFlyer reports India accounts for nearly 28% of all global Android install fraud, and that the affiliate share of India's fraud grew from 57% to 65% in a year [11]. With e-retail media the fastest-growing line of Indian digital spend [12], the inventory types that run hottest for fraud elsewhere are the ones India is buying more of.
The surrounding fraud economy is visible in government data. Financial-fraud complaints on India's National Cyber Crime Reporting Portal rose from 6,94,446 in 2022 to 24,02,579 in 2025, with reported losses of Rs 22,495 crore in 2025, according to a Ministry of Home Affairs written reply in the Rajya Sabha in February 2026 [30]. The rented devices and mule accounts behind consumer scams are the same resources a click-farm operator rents, so an Indian advertiser should assume the supply is local and plentiful.
Why does invalid traffic still reach advertisers?
Because the platforms filter what they can prove, and the rest lands on your account. TAG's European analysis found 24% of spend still ran through unprotected channels at more than 10 times the relative loss of protected ones [2], and the ANA's Q4 2025 benchmark found that fewer than half of participating advertisers, 43.8%, cleared its 50% quality threshold [19].
The supply keeps growing. Cloudflare found about a quarter of global bot traffic coming from two cloud providers [6], and Meta's own estimate that 5% of daily active people are fake accounts shows what survives the removal of a billion accounts in a quarter [9].
- Timing: platform filtering runs at auction and billing time, and Google's 60-day investigation window closes before most quarterly reconciliations [25].
- Definition: the platform decides what counts as invalid and reports it only in aggregate, as estimated savings and credited cost [25].
- Scope: Google sees the click and the lead form, but 52% of fraudulent installs now arrive through organic, where no ad platform looks [11].
- Coverage: Microsoft Invest states it blocks some cloud providers but not all, and needs 10,000 impressions on a slice of inventory before an anomaly counts as significant [27].
- Scale: with 53% of web traffic automated [1] and bot activity growing 63% in APAC [18], any fixed filter loses ground every quarter.
The question is not whether invalid traffic exists in your account, but whether anyone measures it with the advertiser's interest in mind. Web Protection is built for that side of the table.
How ClickTrust helps
ClickTrust runs 320 detection traps, each labeled GIVT or SIVT in line with MRC invalid-traffic guidance, and scores every event in under 50 milliseconds into one fraud score, the Traffic Quality Index (TQI). IP and audience exclusions are pushed to Google Ads, Meta and Bing automatically, so the rate you measure starts to fall the day you measure it.
Three layers share one workspace. Web Protection covers search, social, display, affiliate and direct traffic through a JavaScript tag. In-App Protection ships native iOS, Android and Flutter SDKs for emulators, click injection, repackaged apps and install hijacking. Lite-MMP Attribution adds tracking links, server-to-server postbacks and a fraud score per affiliate. Every flagged event feeds a refund-ready evidence pack for the Click Quality Form, and an MCP server lets AI agents query the same data.
ClickTrust is free to start. If you want a number before you install anything, Traffic Rx gives a free landing-page traffic diagnosis.
What advertisers should do
The playbook
01
Measure before you protect
Run a monitor-only baseline for one billing cycle on every paid channel. TAG's certified-channel benchmark is 1% [8] and its unprotected European rate is 10.36% [2]. You cannot claim a credit or shift budget without your own number.
02
Move spend into certified channels
The 24% of European spend outside protected channels carried more than 10 times the relative loss [2]. Ask every agency, DSP and SSP in your chain for its Certified Against Fraud status, and treat private marketplaces as the default, as ANA participants do with more than 92% of median spend [19].
03
Read the Google card monthly
Check estimated savings and credited cost in the invalid traffic card, pull the Invalid Activity Credit Report by campaign and network, and file the Click Quality Form inside the 60-day window with evidence [25, 26].
04
Block data-center and cloud traffic first
About a quarter of global bot traffic comes from AWS and Google Cloud alone [6], and Microsoft Invest states it blocks some cloud providers but not all [27]. Data-center IP exclusions are the cheapest filter you control.
05
Score affiliates and installs separately
Affiliates run near a 40% install-fraud rate against about 1% for self-reporting networks, and organic carries 52% of fraudulent installs [11]. Per-partner fraud scores and SDK-level checks catch what click filtering cannot.
06
Audit lead quality, not only click quality
Google now investigates poor-quality form submissions and says it cut invalid-lead exposure by 30% in 2026 [26, 24]. Score every form submission against the session that produced it and reject fake-account patterns before they reach sales.
Methodology
How this report was built
- This report synthesizes 30 published sources from standards and industry bodies (MRC, TAG, ANA), ad platforms (Google, Meta, LinkedIn, Microsoft), analysts (Juniper Research, dentsu, FICCI-EY, WPP Media), infrastructure and security companies (Imperva, Cloudflare, Fastly, Akamai), the mobile measurement partner AppsFlyer and the Government of India. Every figure is attributed in the sentence where it appears.
- It does not cite ad-verification or click-fraud vendors, and no figure in it originates from one. Where TAG's European report blends data from accredited measurement vendors, the text says so and attributes the figure to TAG.
- Figures were checked against the publisher's own page or PDF wherever one was available. Where only a trade-press report could be read (Social Samosa for dentsu and WPP Media, PPC Land for the WFA estimate), the sentence says so and the source entry names the outlet.
- Measurement models are not interchangeable. Modeled cost estimates, network-edge bot shares, certified-channel benchmarks and platform transparency counts are presented side by side and never averaged.
- Data periods range from 2023 (Juniper Research, the ANA's first study, TAG's European data) to the third quarter of 2026. Older figures are labeled with their year in the text, and the Juniper cost estimate is reported as a 2023 estimate throughout.
- ClickTrust's own data appears only as the audit range of 14 to 42 percent invalid traffic depending on channel, which comes from accounts that sought an audit and is therefore not a market average. Currency figures are reported as published without conversion.
Sources
References
- [1]Imperva (a Thales company), Bad Bot Report 2026: The Internet Is No Longer Human and It's Changing How Business Works (April 2026)
- [2]Trustworthy Accountability Group (TAG), 2025 European Ad Fraud Savings Report (executive summary PDF) (2025 (2023 data))
- [3]Juniper Research, Quantifying the Cost of Ad Fraud: 2023-2028 (whitepaper) (2023)
- [4]Google, 2025 Ads Safety Report (PDF) (April 2026)
- [5]Association of National Advertisers (ANA), The ANA Q2 2025 Programmatic Transparency Benchmark Finds $26.8B in Wasted Programmatic Spend (August 2025)
- [6]Cloudflare, The 2025 Cloudflare Radar Year in Review (December 2025)
- [7]Fastly, New Fastly Threat Research Reveals AI crawlers make up almost 80% of AI bot traffic (August 2025)
- [8]Trustworthy Accountability Group (TAG), US Ad Industry Extends Success in Fight Against Fraud with Fourth Straight Year of Sub-1% IVT in TAG Certified Channels (October 2024)
- [9]Meta (filed with the EU Transparency Centre, disinfocode.eu), Code of Practice on Disinformation report, March 2026 (covering July to December 2025) (March 2026)
- [10]LinkedIn, Community Report, H2 2025 edition (Transparency Center) (2026)
- [11]AppsFlyer, The State of Fraud for Marketers: 2026 Edition (April 2026)
- [12]Social Samosa (reporting the dentsu-e4m report), Digital becomes India's largest ad channel: dentsu-e4m Digital Advertising Report 2026 (February 2026)
- [13]Media Rating Council, Media Rating Council Issues Update To Invalid Traffic Detection And Filtration Standards (June 2020)
- [14]Google, Google Ad Traffic Quality (Undated, fetched November 2026)
- [15]Media Rating Council (via PR Newswire), Media Rating Council (MRC) Issues Final Version Of Invalid Traffic Detection And Filtration Guidelines Addendum (October 2015)
- [16]PPC Land (citing WFA), Bot traffic costing advertisers billions as fraud detection fails, investigation reveals (quoting the World Federation of Advertisers estimate) (March 2025)
- [17]Juniper Research, The Distillery #22: Network APIs Go Agentic & Inside the Scam Social Media Ads Surge (February 2026)
- [18]Akamai, Akamai Research: Commerce Becomes the Epicenter for AI Bot Attacks and Agentic Fraud in 2026 (July 2026)
- [19]Association of National Advertisers (ANA), ANA Q4 2025 Programmatic Transparency Benchmark Shows Media Quality Is Now Driving Measurable Performance Outcomes (February 2026)
- [20]Association of National Advertisers (ANA), ANA Provides "First Look" at In-depth Programmatic Media Transparency Study (June 2023)
- [21]Meta (filed with the EU Transparency Centre, disinfocode.eu), Report September 2025 (Code of Practice on Disinformation) (September 2025)
- [22]Meta (Transparency Center), Integrity Reports, H1 2026 (March 2026)
- [23]Google India, Gemini is stopping harmful ads before people ever see them (India edition of the 2025 Ads Safety Report) (April 2026)
- [24]Google, We're introducing a new transparency tool for invalid traffic in Google Ads (November 2026)
- [25]Google, About invalid traffic (Google Ads Help) (Undated, fetched November 2026)
- [26]Google, About the Invalid Activity Credit Report (Google Ads Help) (2026)
- [27]Microsoft, Microsoft Invest: Traffic Quality (Microsoft Learn) (March 2025)
- [28]EY India (FICCI-EY), India's media and entertainment sector grew 9% to INR2.78 trillion in 2025 (FICCI-EY report press release) (March 2026)
- [29]Social Samosa (reporting WPP Media's This Year Next Year), India ad revenue to reach Rs 2 lakh crore in 2026: WPP Media TYNY report (February 2026)
- [30]Ministry of Home Affairs / Indian Cyber Crime Coordination Centre (I4C), Cybercrime Reporting and Investigation (written reply in the Rajya Sabha, PIB release) (February 2026)
Frequently asked questions
- What percentage of ad traffic is invalid?
- It depends on the channel and whether it is protected. TAG measured invalid traffic under 1% in certified channels in the US and about 1% in protected European channels, against 10.36% in unprotected European video and display, using 2023 data [8, 2]. Juniper Research's 2023 estimate put 22% of global digital ad spend lost to fraud [3]. ClickTrust audits of individual accounts find 14 to 42 percent depending on channel.
- Does Google refund invalid clicks?
- Yes, for the invalid traffic Google itself detects. Google Ads Help says invalid traffic caught before billing is never charged, and invalid traffic caught afterwards is returned as credited cost [25]. You can request an investigation of the last 60 days through the Click Quality Form, which also covers poor-quality form submissions, and the Invalid Activity Credit Report shows credits by campaign and network [25, 26].
- How much of web traffic is bots?
- More than half, by the broadest measure. Imperva's Bad Bot Report 2026 found automated traffic exceeded 53% of all web traffic in 2025, with humans at 47% [1]. Cloudflare found humans generated only 47% of HTML requests on its protected sites by December 2025 [6], and Fastly measured automated traffic at 37% of activity on its network in mid-2025 [7].
- Is click fraud worse in India than elsewhere?
- No allowed source publishes an India invalid-click rate, so the answer rests on exposure. India accounts for nearly 28% of global Android install fraud per AppsFlyer [11], Google blocked or removed 483.7 million ads and suspended 1.7 million accounts in India in 2025 [23], and digital spend reached Rs 71,621 crore in 2025 per the dentsu-e4m report [12]. ClickTrust audits find 14 to 42 percent invalid depending on channel.
- What is the difference between GIVT and SIVT?
- GIVT, or General Invalid Traffic, is caught by routine and list-based filtering: known bots, spiders, crawlers and non-browser user agents, per the MRC's 2015 guidelines [15]. SIVT, or Sophisticated Invalid Traffic, needs advanced analytics, multipoint corroboration and human review: hijacked devices, adware, malware and misappropriated content [15]. The MRC's 2020 update added more SIVT categories and a required decision-rate metric [13].
- How much does ad fraud cost globally?
- Estimates range by method and year. According to a 2023 estimate from Juniper Research, ad fraud cost $84.2 billion, or 22% of digital spend, and was forecast to reach $172.3 billion by 2028 [3]. The ANA puts programmatic waste at $26.8 billion a year as of Q2 2025 [5], and TAG estimated European anti-fraud standards saved €3.45 billion in 2023 [2].