How to Prepare Evidence for a Google Ads Invalid Traffic Investigation

When an investigation makes sense

Escalate a recurring concern that remains unexplained after checking tracking, campaign changes and ordinary user behaviour. State what Google's billing review should resolve, rather than asking it to validate every sales rejection. Preserve original records before changing controls and distinguish already filtered activity from the unresolved concern.

Complete the first-party investigation before preparing an escalation

Google's current 60-day consideration window

As checked on 7 October 2026, Google accepts invalid-traffic investigation requests for activity within the last 60 days. This is a provider review window, not an advertiser data-retention recommendation or a promise that every case receives a credit. Verify the current process before submission and make the affected dates explicit.

Customer ID and affected campaign

Include the Google Ads Customer ID and the affected campaigns, ad groups and keywords where relevant. Keep stable identifiers alongside readable names; names may change during an investigation. If the available records cannot attribute a lead to a particular ad group or placement, describe that gap instead of supplying a guessed value.

Date range

Identify the start and end of the suspected activity, the timezone and when the evidence was exported. Keep interaction dates, submission dates and later CRM review dates separate. A comparison period can establish what changed, but should use compatible campaign settings, tracking definitions and follow-up time.

GCLID

Provide the available click identifiers and their association with the affected requests or leads. Preserve the original value and document duplicate or overwritten attribution. A GCLID supports an association with an ad interaction; it does not certify the lead's identity, genuine interest or eventual billing classification. Mark missing associations as unmatched.

IP/network evidence

Attach relevant server observations with timestamps and explain the repeated pattern. A shared address, mobile carrier exit or proxy classification is network context, not proof of one actor. Separate the address visible to your infrastructure from assumptions about the original client, and avoid accusing an organisation from an IP match alone.

Interpret shared and changing addresses before proposing an exclusion

User agent

Retain observed browser or device information where available. Explain how it relates to the suspected sequence and whether records came from the server or client analytics. A User-Agent string is a declaration and can be altered; it does not independently establish a bot, person or operator. Missing fields should stay missing.

Referrer

A referrer can add navigation context when your logs contain it. Browsers, privacy settings and redirects can omit or shorten the value. Treat it as optional supporting evidence: the current Google checklist does not make a referrer a mandatory field, and an absent header is not a fraud verdict. Do not infer an exact placement from a generic referrer.

Lead quality/outcome evidence

For lead-related concerns, Google asks for click identifiers and the advertiser's assessment of leads, including results of contact attempts. Use documented outcomes such as invalid number, denied enquiry, genuine but unsuitable, or pending. One unanswered call is not confirmed fabrication. Share only the evidence needed through the approved support process.

Use consistent reasons for rejecting or qualifying a lead

Building a concise chronology

A reviewer should be able to trace the concern without reconstructing your entire account. Separate observed events from your interpretation and keep source references for representative cases. Include legitimate alternatives and measurement changes that could explain the pattern.

• Identify the campaign, period and business symptom.
• Describe the comparison cohort and evidence coverage.
• Reference a limited set of representative request/lead records.
• List the hypothesis, uncertainty and checks already performed.
• State the question requiring Google's review, without prescribing a refund.

Interpret any later credit separately from lead-quality findings

What the evidence cannot prove

An independent evidence set does not expose Google's confidential per-click classification or guarantee recovered spend. A repeated technical pattern can justify an operational response without proving who caused it. AdFraud.pl does not submit Google investigation requests or recover advertising credits automatically on the basis of this article.

Scope a review of the evidence before escalating a billing concern