Ad-fraud protection for agencies and EC businesses that care about every yen of ad spend

Bots are quietly eating your ad budget.
Detect them. Stop them. Prove it.

Disguised bots that IP and User-Agent blacklists cannot stop are caught by browser fingerprinting and behavioral signals. Exclusion signals go to TikTok, Meta and Yahoo! ads to protect both your budget and your campaigns’ learning data. Every verdict comes with its reasons — no black box.

Free plan: no credit card · 10,000 PV/mo free · Detection live in 5 minutes.

Paid plans (Starter / Pro / Business) include a 14-day free trial. A credit card is required to start, and the plan automatically converts to paid when the trial ends.

shop.example.jp

Last 24 hours

Demo data

Human

1,104

Suspicious

113

Bot

234

IP addressVerdict
203.0.113.42Bot88
203.0.113.17Human4
203.0.113.91Bot76
203.0.113.28Human0
203.0.113.63Suspicious49

From the click to the ad platform

  1. 01

    An ad is clicked

    The tag on your landing page loads and sends details about that click.

  2. 02

    We judge it on the spot

    Browser fingerprint, IP and behavioural signals decide whether it is a person or a bot.

  3. 03

    We decide whether to act

    Even when we call it a bot, nothing is sent unless it clears the threshold you set for that site.

  4. 04

    We signal the ad platform

    Only above the threshold do we send an exclusion signal to TikTok, Meta or Yahoo! JAPAN.

Detection and action use separate thresholds, so a person is not excluded by mistake. The exclusion signal builds an audience in your ad manager; delivery only stops once you exclude that audience yourself.

What we publish

4

Ad platforms

TikTok, Meta, Yahoo! JAPAN and Google Ads.

7 days

Raw log retention

Records containing IP and user agent are deleted automatically after seven days.

Every visit

Reason shown

You can see why each individual visit was judged the way it was.

Detection logic — open, not a black box

AdFraud Shield discloses *why* a click was flagged as a bot. That's the line that separates us from legacy tools that lean on User-Agent strings and IP blacklists.

Browser

Canvas & WebGL render fingerprint

Headless environments without a real GPU produce SwiftShader-specific pixels. Missing WebGL extensions are a separate bot signal.

Interaction

Behavioral honeypots

Invisible DOM links and form fields that no real user reaches. Touching one is among the highest-confidence bot signals we have. Scrapers fall in by structure.

Identity

Geo & timezone reconciliation

Compare the JS-reported browser timezone to the IP geolocation. The moment a datacenter IP claims Tokyo, we know.

Network

TCP/IP TTL hop analysis (edge-dependent)

Reverse-infer the OS from the inbound packet's initial TTL. Engine-implemented; it requires a network-edge layer and is not enabled in the current configuration.

Protocol

TLS JA4 fingerprint (edge-dependent)

Generate a JA4 hash from the ClientHello — cipher suite, extensions, elliptic-curve order. Engine-implemented; it requires a network-edge layer and is not enabled in the current configuration.

Protocol

HTTP/2 SETTINGS frame (edge-dependent)

HTTP/2 SETTINGS ordering and WINDOW_UPDATE timing are an implementation-specific fingerprint. Engine-implemented; it requires a network-edge layer and is not enabled in the current configuration.

The current options — and where we sit

There's been a gap between cheap-but-shallow overseas tools and expensive long-contract enterprise products. We fill it.

Generic overseas tools

  • IP blacklist + User-Agent only
  • Miss residential proxies and device-fingerprint spoofing
  • No Japanese support, no Yahoo! JAPAN integration

Cheap. Insufficient for the Japanese market.

Legacy enterprise products

  • Annual contracts in the millions of yen
  • Months of onboarding, dedicated CS expected
  • Weak fit for agencies juggling many sites

Built for Fortune 500. Overkill for everyone else.

AdFraud Shield

  • Multi-layer analysis: Canvas/WebGL fingerprint, behavior, datacenter IP
  • From ¥9,800/mo · zero setup fee · no minimum term
  • Native sync for Google · TikTok · Meta · Yahoo! JAPAN

Built for agencies and mid-market e-commerce in Japan.

Works with your existing ad stack

Native support for Google · TikTok · Meta · Yahoo! JAPAN. Drop the JS tag from GTM, Shopify, or anywhere.

Google AdsTikTok AdsMeta AdsYahoo! JAPANGoogle Tag ManagerShopifyWordPress
JS tag
<script async src="https://storage.googleapis.com/adfraud-shield-jstag/v1/adfraud-shield.min.js"
        data-site-id="YOUR_SITE_ID"
        data-endpoint="https://adfraud-shield.io/api/v1/analyze"></script>

GTM: Custom HTML tag, fire on All Pages.

Simple, transparent pricing

A flat monthly base fee with a generous included-PV allowance, then a low per-PV rate only above it (overage never stops protection). Business covers up to 30 sites for agencies.

Free

See your bot exposure first

¥0/mo
  • 10,000 PV / month
  • 1 site, 1 user
  • Bot detection report
  • Dashboard basics
Start free

Starter

Single store or small agency

¥9,800/mo
  • 300,000 PV / month
  • 3 sites, 3 users
  • Google Ads auto IP exclusion
  • CSV export + monthly Traffic Quality Report (PDF)
Choose Starter
Popular

Pro

Mid-market e-commerce, multi-channel

¥29,800/mo
  • 1,000,000 PV / month
  • 10 sites, 5 users
  • TikTok / Meta / Yahoo! JAPAN integrations
  • Weekly/monthly reports + Protection Wiring Drill
Choose Pro

Business

Agencies and large e-commerce

¥69,800/mo
  • 5,000,000 PV / month
  • 30 sites, 10 users
  • All integrations + priority support
  • Cross-site portfolio dashboard
Choose Business

Enterprise — beyond 10M PV/mo · custom integrations · SLA · dedicated CSM

Contact us

Questions, answered

Make today the last day you pay for bots.

Start on the free plan and see how many bots are passing through your site. Decide after you have seen the numbers.

Start measuring (free)

Free plan covers 10,000 PV/month. No credit card required.

AdFraud Shield is operated by FID Ltd., which holds Japan's PrivacyMark and is certified to ISO/IEC 27001:2022.

View registration details