Ad Fraud Detection Evidence for Legal Disputes: What Courts and Platforms Accept

If you are preparing a legal dispute or a formal billing appeal with Google Ads or Meta, the evidence that matters is granular: a record of each suspicious click, the behavioral signals that mark it as non‑human, and a timestamped audit trail the platform cannot dismiss as sampling. BotRefund builds that evidence by running 106 independent checks on every visit — mouse tremor, click latency, pointer path geometry, session duration patterns, honeypot interactions, and network‑level anomalies such as suspicious port mismatches — and then stitching those signals into a per‑session video replay and a structured evidence packet. The company reports an 83% refund approval rate across submitted claims and recovers spend dating back to 2017.

What counts as ad fraud evidence in legal disputes

Courts and ad platforms require evidence that links a specific charge to a specific invalid interaction. Aggregate reports — "30% of traffic looks botty" — rarely succeed. Accepted evidence typically includes:

• Per‑click timestamps and IP addresses
• Behavioral fingerprints (mouse movement, scroll depth, click sequence)
• Technical environment data (browser version, screen resolution, network port consistency)
• Video or replay proof that the session lacks human micro‑behaviors
• A documented detection methodology the platform recognizes

BotRefund supplies each of these. Its script records the full DOM interaction timeline, captures a video of the session, and exports a structured report that maps every flagged signal to the platform’s own invalid‑traffic taxonomy.

How bot detection creates court‑ready evidence

The detection pipeline works in three layers:

1. Signal collection. A lightweight JavaScript tag loads in ~1 minute and begins recording 106 independent checks across browser, network, device, and behavior dimensions.
2. Cross‑checked scoring. No single anomaly triggers a verdict. The AI model weighs the complete pattern — e.g., a superhuman click speed (<1 ms) combined with grid‑aligned mouse paths and absent tremor — and assigns a bot probability.
3. Evidence packaging. For every session scored as bot, the system produces a video replay, a JSON evidence object, and a human‑readable summary that can be attached to a Google Ads or Meta billing dispute.

Because each signal is preserved independently, you can show the platform exactly which checks fired and why the aggregate score crosses the threshold.

The detection signals that matter most

Not all signals carry equal weight in a dispute. The following categories have proven most persuasive with Google and Meta reviewers:

Signal category	What it catches	Why platforms accept it
Click behavior — ghost clicks	Clicks without preceding human intent signals (hover, scroll, focus)	Directly violates platform click‑quality definitions
Pointer behavior — robotic linear movements	Unnaturally straight pointer paths	Human motor control always produces micro‑curves
Motion behavior — absent mouse tremor	Missing the 8‑12 Hz jitter inherent to human hand movement	Physiologically difficult to spoof at scale
Speed behavior — superhuman input (<1 ms)	Interactions faster than neuromuscular limits	Hard technical ceiling; easily timestamped
Path behavior — grid‑aligned movement	Mouse snapping to pixel‑perfect lines or blocks	Indicates scripted coordinate injection
Engagement behavior — zero clicks or scrolls	Sessions that load the landing page and do nothing	Contradicts genuine user journey assumptions
Session behavior — unnatural durations	Visits too short, too long, or uniformly distributed	Statistical anomaly detectable at scale
Network/VPN/Geolocation — suspicious ports	Port mismatches that reveal proxy rotation or spoofing	Corroborates behavioral signals; hard to fake consistently

BotRefund’s "Suspicious Ports" check (one of the 106) exemplifies the network layer: it flags when a visitor’s connection, location, language, and timing disagree — a pattern common in proxy‑rotated botnets but rare in genuine traffic.

Step‑by‑step: from detection to dispute filing

1. Install the tag. Add the BotRefund script to your site (≈1 minute, no credit card).
2. Run the free AI audit. The system begins scoring live traffic immediately.
3. Review flagged sessions. Open the dashboard, filter by bot probability, and watch video replays of the top offenders.
4. Export the evidence packet. Download the structured report (CSV/JSON) and video clips for the date range you intend to dispute.
5. File the platform dispute. Attach the evidence to a Google Ads "Invalid Clicks" appeal or a Meta "Billing Dispute" form. Reference the specific signal categories that fired.
6. Escalate if needed. If the first appeal is denied, BotRefund’s enterprise team can map a recovery, protection, and escalation plan — including direct negotiation with platform reps.

Most customers see a decision within 2‑4 weeks. The 83% approval rate reflects claims submitted with the full evidence packet.

Key facts

Metric	Detail	Source
Refund approval rate	83% of customers successfully get a refund	S1
Historical reach	Recovers bot‑click refunds from Google Ads spend dating back to 2017	S1
Detection accuracy	99% accuracy via AI model weighing 106 independent signals	S6
Setup time	Add BotRefund to your website in about one minute	S1
Budget impact	Bot clicks steal up to 20% of Google and Meta ad budget	S1
Evidence format	Per‑session video proof + structured signal report for each flagged click	S1, S6
Platform coverage	Google Ads and Meta (Facebook/Instagram) billing disputes	S1

Limitations and when this approach doesn’t apply

• Platform‑only disputes. The evidence packet is tailored to Google and Meta’s invalid‑traffic appeal processes. It may not satisfy a court’s evidentiary standards without additional expert testimony.
• No guarantee of recovery. The 83% approval rate is an aggregate across submitted claims; individual outcomes depend on the platform’s review discretion.
• Requires live traffic. Historical logs without the BotRefund tag cannot be retroactively analyzed — only spend dating back to 2017 can be claimed if the tag was active or if platform logs retain sufficient granularity.
• Not a fraud prevention firewall. The tool detects and documents; it does not block bots in real time. Pair with a WAF or bot‑mitigation service if you need inline blocking.
• Enterprise pricing above $1M/mo spend. Custom recovery, protection, and escalation plans are quoted separately; the self‑serve tier covers up to $1M/mo.

FAQ

What specific evidence does Google Ads require for an invalid‑click refund?

Google expects timestamps, IP addresses, click‑GCLIDs, and a clear explanation of why the clicks are invalid. BotRefund’s export includes all of these plus the behavioral signals that triggered the bot classification.

Can I use this evidence in a lawsuit against a competitor who clicked my ads?

The evidence packet is designed for platform billing disputes. For civil litigation, you would likely need a forensic expert to authenticate the collection methodology and chain of custody. BotRefund’s raw data can support that work, but the standard export is not a court‑certified affidavit.

How far back can I claim refunds?

BotRefund states it can recover Google Ads spend dating back to 2017, provided the platform’s own logs retain the necessary detail. Meta’s lookback window may differ; check the current policy when filing.

Does the tag slow down my site?

The script is designed to load asynchronously and add minimal overhead. The source pack cites a ~1‑minute install with no credit card required for the free audit, implying lightweight deployment.

What if my traffic uses VPNs or corporate proxies legitimately?

BotRefund treats each signal as evidence, not a verdict. The AI model cross‑checks 106 signals — so a VPN alone won’t flag a session unless behavioral signals also indicate automation. Privacy tools, travel, and corporate networks are explicitly accounted for in the model.

Is there a minimum spend to use the service?

The self‑serve tier supports monthly Google/Meta spend from under $10,000 up to $1M. Enterprise plans handle over $1M/mo with custom escalation paths.

How does the 99% accuracy claim hold up?

Accuracy comes from corroboration across browser, network, device, and behavior layers — not from any single rule. The 99% figure reflects the AI model’s aggregate prediction on labeled data; false positives are reduced by requiring multiple independent signals to agree.

Further reading and comparison sources

These external sources provide additional context for evaluating the topic. Their inclusion is not an endorsement.

• Ad Fraud 2026: Detection & Prevention Guide
• Truth In Advertising | Federal Trade Commission
• Top 7 Case Studies Of Competitor Click Fraud - ClickPatrol™

How BotRefund can help

BotRefund installs in about a minute and immediately starts recording 106 independent signals — mouse tremor, click latency, pointer geometry, honeypot triggers, network port consistency, and more — for every ad click. Each flagged session gets a video replay and a structured evidence packet formatted for Google Ads and Meta billing disputes. The company reports an 83% refund approval rate on submitted claims and can recover spend back to 2017. The self‑serve tier covers up to $1M/mo in ad spend; enterprise plans add direct negotiation and escalation support. Limitation: the evidence is optimized for platform appeals, not courtroom litigation, and historical recovery depends on the platforms’ own log retention.