If you think a competitor or a bot is draining your budget, first check whether you paid for those clicks at all. Google filters most invalid clicks before it bills you, and it credits some of the rest later. What's left is what an investigation request is for, and Google asks for specific evidence.
What Google does on its own
Google uses automated filters and manual reviews to find invalid traffic: bots, accidental clicks, fraudulent placements, and people with no interest in your business, such as a competitor clicking to raise your spend. What happens next depends on when it is found:
| Found | What Google does | Where you see it |
|---|---|---|
| Before the end of the billing cycle | Removes it from your bill and your campaign metrics, and returns the cost to your campaign budget | The Invalid clicks column (Campaigns → Columns → search for "Invalid clicks") |
| After you were billed | Credits it in the next billing cycle, when appropriate and possible | Billing → Summary → the Month card → Adjustments, labeled Invalid activity; and the Invalid Activity Credit Report in the Report Editor |
The two numbers aren't directly related. Many filtered clicks and a small credit is normal: the column counts clicks you never paid for, and the credit covers what Google found after the bill closed.
Source: Google Ads Help, About invalid traffic.
Rule out the ordinary first
Google's own FAQ lists the patterns that look like fraud and often aren't:
- A traffic spike. Seasonal demand, a competitor pausing their ads or your own budget, keyword or location changes can all cause one. Check your change history.
- Zero-second visits. Many analytics tools need a second click to measure a visit, so a visitor who read your landing page and left can show as zero seconds.
- Repeated IP addresses. Offices, universities, cafés and mobile carriers put hundreds or thousands of devices behind one address.
- Clicks from outside your targeted area. Someone searching for your location by name, or on that country's Google domain, can see your ad.
Also look at the Invalid clicks column before you build a case. A third-party tool can flag a click that Google already filtered, and that click was never on your bill.
When to ask for a review
Ask when the pattern survives those checks: the same sources clicking again and again with no conversions, a spike with no matching rise in conversions, or leads that turn out to be fake when you call them. Google investigates activity from the last 60 days, so don't wait for the month to close.
What to send
Google asks you to gather this before you submit:
- Your 10-digit Customer ID.
- The exact dates of the suspicious activity.
- The campaign and ad group names, and any keywords driving the traffic.
- Evidence from your web server logs: the IP addresses behind repeated or non-converting clicks, the browsers and devices involved, and the GCLID of each suspicious click, which lets Google match the visit to the ad interaction.
- For invalid leads, a spreadsheet with each lead's GCLID and your assessment, such as what happened when you emailed or called.
- A short explanation of why the traffic is suspicious, for example clicks up 300% with no more conversions, or clicks from outside your targeted locations.
How to submit, and what comes back
Send it through Google's Click Quality Form. Specialists review the last 60 days of your traffic, which usually takes several business days, and you get an email with their findings. If they find invalid traffic the filters missed, Google tells you about the credit.
Don't expect a click-by-click answer. Google doesn't confirm which specific clicks it judged valid or invalid, because that would show bad actors how its detection works.
A flag is evidence, not a verdict
Your logs, and any tool's verdicts, are evidence for Google's review. Google decides what counts as invalid and whether a credit is due. Nothing guarantees a refund.
Where AdProtektor fits
The evidence Google asks for comes from your own site, which is where AdProtektor works. Its Google Ads Refund Report, available on every plan, exports a CSV for the dates you choose:
- Every flagged ad click with its GCLID, click time in UTC and in your time zone, IP address, verdict (such as bot, competitor or click farm), threat score, time on page, scroll depth, interactions and campaign.
- A summary per IP address, with its flagged clicks and their GCLIDs.
- Flagged clicks and the flagged rate per campaign, plus engagement anomalies.
It can't see which of those clicks Google already filtered, so compare the dates with your Invalid clicks column before you file. And a credit only covers clicks already made. To stop the same sources coming back, AdProtektor blocks them and writes their addresses to your Google Ads IP exclusions; our guide to the IP exclusion limits explains how those lists fill up.
Google pays back what it agrees was invalid. Keeping the same sources away from your ads is your side of the job.
To see which clicks on your ads look invalid, start a free trial and add the tracking script to your site.