Best IVT detection
17 min read
2025 broke the assumption that MRC accreditation means a vendor reliably blocks bots…
Simul Sarker
CEO of DataCops
Last Updated
May 10, 2026
Best IVT detection in 2026: a three-layer breakdown of who actually catches the bots
2025 broke the assumption that MRC accreditation means a vendor reliably blocks bots. The Adalytics 240-page report, published March 28 2025, found that Integral Ad Science labeled known URLScan.io bot traffic as human 77 percent of the time across a 2019 to 2024 dataset. DoubleVerify missed the same declared bots 21 percent of the time. The DV stock dropped 36 percent in a single day on February 28 2025, falling from 21.73 dollars to 13.90 dollars, and a securities class action covering the November 2023 to February 2025 window followed in July. Stock is now down roughly 70 percent from its peak.
Meanwhile the IVT rate did not improve. Fraudlogix's 2026 report put global IVT at 20.64 percent across 105.7 billion impressions analyzed. Pixalate's Q1 2026 benchmarks: 20 percent on web, 39 percent on mobile app, 25 percent on CTV across 82 billion impressions. Juniper Research projected 100 billion dollars in global ad-fraud losses for 2026, scaling to 133 billion dollars by 2028, driven by AI botnets and autonomous agents. About 21 percent of programmatic impressions now come from made-for-advertising sites, often hiding inside Performance Max where the buyer cannot inspect them.
The single biggest mistake in this category right now is shopping for an IVT vendor without understanding which layer it operates at. Pre-bid blocks ad serving before the impression. Click-time stops a Google Ads click after the auction but before the form view. Conversion-time validates the actual conversion event before it ever reaches your CAPI and Meta's Andromeda or Google's Smart Bidding training set. These are different vendors. Most listicles mash them together, score them on feature counts, and miss the question that decides everything: which layer matters for your campaign?
This post breaks the field into the three layers and tells you honestly which vendor wins which lane.
Quick stuff people keep asking
What is IVT and how is it different from a bot?
IVT means Invalid Traffic. The MRC defines two flavors: GIVT (general, declared bots, datacenter IPs, anything you can identify from a list) and SIVT (sophisticated, residential proxies, automation tools that mimic human behavior, hijacked devices). GIVT is what most filters catch. SIVT is what gets through everything and trains your bid algorithms on garbage.
Is MRC accreditation enough?
No. The 2025 Adalytics report and the DV lawsuit settled that question. Accreditation is a process audit, not an outcome guarantee. Necessary, not sufficient.
Where does conversion-time IVT detection sit in the stack?
After the click and after the form, before the event hits your server-side CAPI. This is the layer that actually protects Smart Bidding and Meta Andromeda from training on bot conversions. Almost no MRC giant operates here. The performance-marketing tools (ClickCease, ClickGUARD, TrafficGuard, Lunio) operate at click-time, which is earlier in the funnel.
What is agentic AI traffic and is it the same as IVT?
It is the new SIVT. Mid-2026 sees real agentic traffic from OpenAI Atlas, Claude for Chrome, and AWS AgentCore showing up on retail and SaaS sites. HUMAN's AgenticTrust dashboard, launched late 2025, surfaces it. The category is evolving from "is this a human or a bot" to "is this a trusted agent with consent or a spoofed bot pretending to be one".
Should I just use Google's invalid clicks credit?
It is something. It is not a defense. Google credits invalid clicks after the fact and only catches a fraction. Independent measurement is what gives you accountability against the platform itself, which is the entire reason this category exists.
Layer 1: pre-bid IVT (the brand-side, MRC-accredited stack)
This is where ads decide not to serve in the first place. Built for big-brand advertisers running RFP-style media buys through DV360, The Trade Desk, and Amazon DSP. The MRC giants live here. So does the Adalytics controversy.
1. DoubleVerify
The Good: MRC-accredited across pre-bid avoidance, viewability, and IVT measurement, the industry baseline for reporting. Native integrations with every major DSP and SSP. Brand suitability plus viewability plus IVT in one stack.
Frustrations: The Adalytics report (March 28 2025) alleged DV billed customers for impressions served to declared bots from known datacenter IPs. Stock dropped 36 percent in a day on February 28 2025. Securities class action filed June 2025. AI-related disclosure suit added December 2025. April 2025 pre-bid rate-card increase happened in the middle of the credibility crisis. Pricing is opaque CPM-based and typically passed through agency fees, so SMBs effectively pay a middleman tax.
Wish List: Public transparent rate card with measurable IVT detection benchmarks. Pre-bid plus post-bid reconciliation that matches third-party log analysis, which is the core Adalytics dispute.
Value for Money: 6/10. Still the default agency-grade ad-verification stack, but the 2025 lawsuit and stock crash put a permanent asterisk next to its IVT-detection claims.
Pricing: CPM-based, opaque, typical buys 50K dollars plus minimums, quoted via sales.
2. Integral Ad Science
The Good: MRC-accredited across viewability, IVT, brand safety. Pre-bid solutions integrate with most major DSPs. Simpler UI than DoubleVerify per peer reviews. AI-driven low-quality AI content blocker shipped beta in 2025, early on the made-for-AI inventory problem.
Frustrations: The Adalytics report found IAS labeled known URLScan.io bot traffic as human 77 percent of the time across the 2019 to 2024 dataset. An ex-employee alleged detection code ran on only 50 percent of impressions. March 2025 securities class action over alleged false statements about pricing pressure. September 2025: agreed to be acquired by Novacap for 1.9 billion dollars, going private, which creates roadmap uncertainty.
Wish List: Transparency on how decisions are made when IVT slips through pre-bid filters. SMB-tier pricing, the floor is too high for performance shops.
Value for Money: 6.5/10. Brand-side ad verification standard, dragged by lawsuits and an ongoing take-private.
Pricing: Custom enterprise pricing only, not published. Cost is central to the 2025 class-action complaint.
The Good: Verifies 20 trillion plus digital interactions weekly across 500 plus global brands, the largest known fraud-signal pool in the category. Top scores on all 9 criteria in the Forrester Wave Q3 2024. Q4 2025 launched AgenticTrust plus AI Traffic Over Time and AI Agent Activity dashboards, adding OpenAI Atlas, Claude for Chrome, AgentCore, and Rye to the detected-agents list. April 2026 earned MRC accreditation for viewability with IVT filtering. Named G2 Winter 2026 Bot Detection leader. Raised 50 million dollars plus in October 2024 from WestCap, Goldman Sachs, ClearSky.
Frustrations: Pricing is enterprise-only and reportedly surges unpredictably with traffic spikes. Dashboard usability is inconsistent, "compelling but not user-friendly" is a recurring G2 theme. Documentation lags product velocity. Effectively zero SMB presence.
Wish List: Predictable pricing tier that does not spike during traffic surges. Documentation that keeps pace with releases.
Value for Money: 8/10. Category leader for enterprise bot and fraud defense, the safe pick if your budget starts with a six-figure number.
Pricing: Custom enterprise pricing only, no public tiers, AWS Marketplace listings exist.
4. Pixalate
The Good: Strongest CTV and mobile-app IVT coverage in the category. Q1 2026 globals: 20 percent web, 39 percent mobile app, 25 percent CTV across 82 billion impressions. MRC-accredited for SIVT detection on desktop and mobile web. Seller Trust Index 2.0 ranks 20 plus CTV SSPs by arbitrage and fraud risk.
Frustrations: Pricing not publicly disclosed, mid-market buyers report feeling out of budget after sales conversations. Heavily ad-tech focused, not a fit for first-party site analytics or e-commerce fraud. Reports skew toward research output, some buyers want more programmatic blocking automation.
Wish List: Published pricing tiers for mid-market buyers. Stronger pre-bid blocking automation rather than primarily report-driven workflows.
Value for Money: 7/10. If you live in CTV or mobile programmatic, hard to beat. Wrong shape of tool for performance marketers.
Pricing: Custom-quote only, targets ad-tech buyers.
5. Anura
The Good: Claims 99 percent plus ad-fraud detection accuracy, reviewers say it largely lives up to the claim. Unlimited free support via email, chat, phone, plus monthly training. Per-request usage pricing scales cleanly with traffic. Reviewers report annual cost paid back via saved PPC waste within 90 days.
Frustrations: Pricing fully gated, no public tiers. Multiple G2 and Capterra reviewers describe Anura as expensive. Less visible to SMB advertisers vs ClickCease and CHEQ. API documentation thinner than enterprise competitors.
Wish List: Published pricing or transparent self-serve tier. Native one-click connectors to Google, Meta, Microsoft Ads.
Value for Money: 7.5/10. If you run high-volume affiliate or lead-gen, accuracy pays for itself. Not the obvious pick for a Shopify store on 5K dollars per month of Google Ads spend.
Pricing: Hidden, contact sales. Per-request SaaS model with minimum tiers. Free trial available.
Layer 2: bot management and WAF (the security-side stack)
This is the bot-defense layer, originally about credential stuffing and scraping, increasingly relevant for ad-fraud signal too because the IPs overlap.
6. DataDome
The Good: Sub-2ms decisioning at the edge. Processes ~5 trillion signals daily and claims to stop 350 billion plus attacks per year. Forrester Wave Bot Management Leader 2024. Customers include Etsy, PayPal, SoundCloud. Reviewers consistently call out a low false-positive rate vs Imperva. Around 36 million dollars ARR with 10K customers per Latka 2024, rare combo of enterprise credibility and SMB volume.
Frustrations: Cost is the loudest complaint, expensive for smaller teams, bills can spike unpredictably with traffic surges. JS library is prone to race conditions unless loaded extremely early. Minimum project sizes reportedly start around 50K dollars.
Wish List: Predictable pricing tier or per-endpoint plan. Lighter-weight client SDK resilient to async loader race conditions.
Value for Money: 8/10. Top-tier bot detection if you are enterprise-sized.
Pricing: Custom enterprise pricing, no public tiers, reported 50K dollars plus minimum.
7. Kasada
The Good: Customers report 60 to 95 percent reduction in bad-bot requests after deployment. No CAPTCHAs, invisible client-side challenge keeps real users frictionless. Set-and-forget reputation. Gartner Bot Management mindshare jumped from 0.5 to 4.8 percent year over year (Dec 2025).
Frustrations: Pricing fully gated. Niche bot-only focus, you will buy more tools to round out the stack. Smaller integration ecosystem than Imperva, Akamai, HUMAN. Detection tuning for nuanced gray bots requires sales engineering involvement.
Wish List: Self-serve mid-market tier. Native fraud and account-takeover analytics dashboards.
Value for Money: 7.5/10. Cleanest pick if you only need bot defense and want to ditch CAPTCHAs.
Pricing: Custom-quote only, no public tiers.
Layer 3: click-time IVT (the performance-marketing stack)
This is where most SMBs and DTC brands actually shop. Tools that block invalid clicks on Google Ads, Meta Ads, Microsoft Ads after the auction but before the conversion. The pricing here is published and the trial periods are real.
8. Lunio
The Good: Cross-channel intelligence, an invalid IP detected on one platform is auto-excluded across 15 plus ad platforms (Google, Meta, TikTok, LinkedIn, X, Reddit, Snap, Pinterest). ISO 27001 and SOC 2 certified. Protects 35,000 plus Google Ads accounts across 130 countries. G2 Leader in Click Fraud. 14-day free traffic audit lets buyers see actual IVT savings before signing. 2026 industry benchmark: gaming 18.49 percent IVT, education 14.41 percent, telecom 14.26 percent, real estate 13.61 percent across 2.7 billion clicks.
Frustrations: Pricing starts at around 500 euro per month, pricey for SMB performance marketers. Custom-gated after the audit. UI feels enterprise-flavored to smaller shops. Long contracts and minimum spend gating mid-market access.
Wish List: Self-serve transparent monthly tier under 200 euro. Deeper attribution-model integration including post-conversion fraud signals.
Value for Money: 7.5/10. Strongest mid-market pick for cross-channel click fraud.
Pricing: From around 500 euro per month custom-quoted, 14-day free traffic audit before commit.
9. ClickCease (now CHEQ Essentials)
The Good: Most popular SMB click-fraud tool by raw customer count, claimed 14,000 plus customers. Direct integrations with Google Ads, Meta, Microsoft Ads. Now backed by CHEQ enterprise tech post-2023 acquisition. 7-day free trial.
Frustrations: Top Trustpilot complaint is the 12-month annual lock-in hidden in small text on the pricing page. Cancel mid-term and billing continues monthly until end of contract. Month-to-month pricing is 30 percent plus higher than the headline annual-billed figure (84 / 104 / 124 vs 63 / 78 / 93 dollars per month).
Wish List: Real cancel-anytime billing. Clearer disclosure of annual lock-in.
Value for Money: 6/10. Solid detection, big customer base, the pricing presentation has burned enough users to read the contract before signing.
Pricing: 99 to 349 dollars per month per G2. Public site shows 63 / 78 / 93 dollars per month annual-billed. 12-month commitment.
10. ClickGUARD
The Good: October 2025 rebrand shipped a redesigned dashboard plus AI cross-channel reporting (Google, Meta, Microsoft Ads). Granular click-rule engine, power users prefer this over ClickCease's automation. No long-term contract, cancel anytime.
Frustrations: Entry pricing jumped post-rebrand. Lite tier caps you at 5K dollars per month of ad spend, most legit advertisers forced to Standard or Pro. Setup complexity higher than ClickCease. Smaller customer base.
Wish List: Self-serve free tier for testing. Native blocking for TikTok and LinkedIn Ads.
Value for Money: 7/10. More sophisticated than ClickCease for power users, expect to land on the 119 to 159 dollar tier.
Pricing: Lite 74 dollars per month, Standard 119, Pro 159. Quarterly and annual discounts. Cancel anytime.
11. ClickPatrol
The Good: Evaluates 800 plus data points per click. Four protection modules cover blocking, remarketing audience cleanup, form-spam in one subscription. G2 4.6, Capterra 4.7, Trustpilot 4.4. EU-headquartered (Netherlands), 7-day free trial, 17 percent annual discount.
Frustrations: Pricing page emphasizes monthly cost but plans are billed annually, top Trustpilot complaint. One Trustpilot reviewer reported a surprise 100 dollar charge during trial. Capped by Google's negative-IP list (limited slots, 30-day rolling expiry) like all click-fraud tools.
Wish List: True monthly billing without annual lock-in. Native Microsoft Ads parity with Google Ads protection.
Value for Money: 7.5/10. Solid mid-market click-fraud tool, do not miss the annual-billing fine print.
Pricing: From 59 euro per month (around 69 dollars) annual-billed.
12. Fraud Blocker
The Good: Cheapest credible entry tier in the category at 69 dollars per month, around 15 percent below comparable competitors. Proprietary fraud-scoring with 100 plus signals per visitor. G2 4.6, Capterra 4.7, Trustpilot 4.4.
Frustrations: AppSumo reviewer flagged it as reactive, only adds negative IPs after the fact. Reports can show wrong fraud metrics, detecting threats on platforms that have been off for months while missing active ones. Same annual-billing-disguised-as-monthly trap as competitors.
Wish List: True real-time pre-click blocking instead of post-hoc IP list maintenance.
Value for Money: 6.5/10. Cheapest legit option, good for SMBs who just want negative-IP automation.
Pricing: From 59 dollars per month annual / 69 dollars per month monthly. 14-day free trial.
13. TrafficGuard
The Good: Processes 1 trillion plus data points monthly. Multi-channel: Google Ads, mobile UA, PPC. Easy setup praised by agencies. Public ASX-listed parent (Adveritas, ASX:AV1) gives transparency on company stability.
Frustrations: Percentage-based pricing (around 2 percent of ad spend) gets ugly above 50K dollars per month. Support frequently criticized as bot-portal-only. Data sometimes does not match Google Ads exactly. Missing native Facebook Ads integration.
Wish List: Native Meta integration. Tiered flat pricing for spenders above 50K per month.
Value for Money: 6.5/10. Solid for sub-50K-per-month advertisers wanting simple click-fraud filtering.
Pricing: Percentage-based around 2 percent of ad spend protected. Free tier available.
14. CHEQ
The Good: Largest IVT and fraud detection player after ClickCease (acquired 2023) and Deduce (acquired Jan 2025) acquisitions. Deduce identity graph covers 185M plus weekly active users and 1.5 billion daily events with 99.5 percent claimed identity-assessment accuracy. Covers paid-traffic IVT plus on-site bot blocking plus lead validation plus AI-generated identity fraud. Trusted by Fortune 500s and major B2C brands.
Frustrations: Pricing fully opaque. Aggressive M&A pace raises product-integration risk, multiple overlapping fraud SKUs to navigate. Heavy implementation lift. Marketing positioning shifted from click fraud to GTM Security to Intelligence Standard for the Human-AI Era in two years, buyers report whiplash.
Wish List: Clearer SKU map between CHEQ Essentials, CHEQ Paradome, and Deduce. Mid-market self-serve plan.
Value for Money: 7.5/10. If enterprise needs end-to-end fraud under one roof, the obvious pick. Budget for sales calls and integration work.
Pricing: Hidden, enterprise contracts. Public-facing SMB lives under ClickCease at 99 to 349 dollars per month.
Layer 4: conversion-time IVT (the missing layer for performance marketers)
This is the layer almost nobody operates at. After the click. After the form. Before the event hits your server-side CAPI to Meta and Google. If a bot makes it through the click-fraud tool and submits a form, the conversion still goes back to Meta and trains Andromeda. Smart Bidding learns from the bot. Performance Max optimizes toward the bot. The fraud cost compounds for weeks.
15. DataCops
The Good: Conversion-time IVT filtering on the same pipeline as first-party CNAME analytics and server-side CAPI. Filters bots, VPNs, proxies, Tor exits before the event reaches Meta CAPI, Google Ads CAPI, TikTok Events API, or LinkedIn Insight CAPI. 350 plus continuous monitoring points. IP reputation database with 361 billion plus IPs and network ranges, 146.4 billion datacenter and cloud IPs, 11.9 billion VPN endpoints, 620 million proxy and anonymizer IPs. CNAME runs on your own subdomain so the filter survives uBlock, Brave Shields, Pi-hole, iOS Safari ITP, and Consent Mode v2.
Frustrations: Brand new compared to HUMAN, DV, IAS. SOC 2 Type II is in progress, not yet active. ISO 27001 planned. Smaller agency-side track record vs ClickCease and Lunio. Not pre-bid, so this is not the tool for big-brand programmatic verification.
Wish List: SOC 2 Type II shipping. ISO 27001. DSAR API with downstream deletion to Meta and Google. SSO and SAML on standard plans. All on the public roadmap.
Value for Money: 8.5/10. The only credible option in the conversion-time layer that bundles tracking, CAPI, consent, and fraud filtering under one bill.
Pricing: Basic free, 2,000 sessions per month, unlimited bot detection. Growth 7.99 dollars per month, 5,000 sessions, unlimited Meta and Google CAPI events. Business 49 dollars per month, 50,000 sessions plus HubSpot. Organization 299 dollars per month, 300,000 sessions. Enterprise: dedicated runtime, dedicated IP DB, custom DPA.
So what should you actually use?
Want enterprise pre-bid with MRC reporting for big-brand programmatic? HUMAN if you can afford it, Pixalate for CTV, DV or IAS if your agency demands it (with the 2025 lawsuits noted).
Want bot management at the WAF or edge layer? DataDome if you can stomach the 50K-dollar floor, Kasada for CAPTCHA-free.
Want click-time fraud filtering on Google Ads with a published price? Lunio for cross-channel mid-market, ClickGUARD for power users, ClickPatrol for EU operators, Fraud Blocker for cheapest credible.
Want SMB-friendly click fraud with a real free trial? ClickPatrol, Fraud Blocker, ClickCease (read the annual-lock fine print).
Want conversion-time IVT that actually keeps Smart Bidding from training on bots? DataCops, then re-evaluate the click-fraud tool above it.
Want one stack covering tracking, CAPI, consent, and fraud? DataCops. None of the others bundle this.
The mistake people make
Buying a click-time fraud tool, watching the dashboard show 14 percent IVT blocked, and assuming the job is done. The conversion-time layer is still wide open. Bots that mimic real form submission still hit your CAPI. Meta Andromeda and Google Smart Bidding still train on the bot conversions. The bid algorithm learns to find more of them. Three months later your CPA looks fine and your real conversions have collapsed.
The other mistake: treating the MRC accreditation badge as proof. The 2025 Adalytics report and the DV securities lawsuit ended that argument. Layer matters. Badge does not.
Related reading:
- DataCops vs Lunio
- DataCops vs Fraud Blocker
- DataCops vs ClickGUARD
- Best Google Ads fraud protection
- Best click fraud protection 2026
Now your turn
Which layer is your stack actually defending? Drop your IVT vendor and which layer it operates at in the comments. If it is one tool, you are probably defending one layer.
