Google’s Invalid Activity Credit Report: What It Refunds, and What It Can’t Undo

What is Google’s Invalid Activity Credit Report?

In June 2026 Google published a help page for a Report Editor template called the Invalid Activity Credit Report. Credits for invalid activity have always appeared as line-item adjustments in your billing and transaction history; the new template breaks those credits down by campaign and by network, and attaches the number of clicks and interactions each credit relates to.

Google explains the mechanism plainly. Its automated defences work in two layers. Most invalid traffic is filtered in real time, before you are billed, and never shows up anywhere. Some activity is identified later, after an invoice has already been issued. For that second category, Google issues a credit “where appropriate and possible”. The report is a window onto that second category only.

That is the entire point of this article, so it is worth saying twice. By definition, every click in this report cleared Google’s real-time filters, served an ad, cost you money, landed on your site, and was only flagged afterwards.

What does the report actually show?

It looks like a normal campaign table with extra columns bolted on. Ten new metrics in two groups:

Notice what the “adjusted” group tells you. Google has to publish a second set of numbers because the first set, the ones in your standard campaign reports and the ones your bid strategy ran on, are not corrected. The original click stays attributed to the month it happened; the credit lands in the following month. Google’s page calls this a “timing mismatch”. Credits issued after a billing cycle closes do not retroactively change campaign data.

How do I pull the Invalid Activity Credit Report?

1. In Google Ads, open the Campaigns menu and go to Report Editor.
2. Scroll to the Template gallery.
3. Select Invalid Activity Credit Report: Search & PMax.
4. Set the date range to a closed billing month (credits for a month arrive in the following month, so last month will be more complete than this one).
5. Segment by network and export. If you run an MCC, check the report is available in each client account; Report Editor access is not always enabled across a manager structure.

Two things you will not find. Any Display, Video, Demand Gen or App campaign, because the template is Search and Performance Max only. And any click older than 60 days, because that is as far back as Google’s click quality investigations go.

What does a Google invalid activity credit refund, and what can’t it undo?

A credit does exactly one thing: it reverses the charge for the click. Everything else the click set in motion is still in motion.

The Smart Bidding row deserves care, because it is the one that costs the most and the one Google is quietest about. Target CPA, target ROAS and Maximize Conversions learn from the clicks and conversions they see. Google does not state that activity it later credits is removed from the signals those strategies train on, and the report’s own adjusted columns exist precisely because the original metrics are never corrected. The reasonable reading is that a refunded click is still a click the model saw. If a bot clicked and “converted” on a form, the model saw a cheap conversion and is now bidding to find more visitors like it. We cover a related version of that mechanism in Can one scammy publisher poison campaign learning?

There is a second piece of evidence, and it comes from Google’s own feature set. Google ships two tools whose only purpose is to pull bad signals out of Smart Bidding by hand: data exclusions, which tell the bidding model to ignore a date range while leaving your reports untouched, and conversion retractions, which permanently remove a conversion and its value from CPA and ROAS bidding. Both exist because reporting data and bidding data are separate pipelines at Google, and because nothing automatic cleans the second one. If credited clicks were purged from the model, Google could simply say so and these tools would be redundant for that case. It has never said so. There is also a timing problem Google documents itself: conversion adjustments must be uploaded within 7 days of the conversion for bidding to take them into account, while a credit for the same click lands the following month. By the time Google tells you a click was invalid, the model’s window for unlearning it has closed.

Why does prevention catch more than Google’s refunds?

Google is not in the cybersecurity business; it is in the auction business, and its job is to protect the auction. Its filters are tuned for a specific question: should this click be billed? A click from a competitor’s office, a scraper, a click farm on residential IPs, an automation framework running a headless browser, or an AI agent browsing on someone’s behalf can all pass that test and still be worthless to you. Google will then, perhaps, catch a portion of them after the invoice and send the money back.

ClickCease asks a different question: should this visitor be allowed to reach your site, fire your pixels and fill your forms at all? It runs on the CHEQ cybersecurity engine, which applies 2,000+ cybersecurity behavioral tests to every visit and blocks invalid visitors in a few milliseconds, before the session registers anywhere. The same engine protects 15,000+ companies at a false positive rate below 0.009%.

The practical difference, step by step:

• Known-bad sources are excluded from the auction. ClickCease pushes IP exclusions and native audience exclusions to Google, Meta and Microsoft Ads, so the ad is not served to them in the first place. Google’s report starts after the ad served.
• Pixel Guard stops the signal. When an invalid visitor does land, Pixel Guard prevents your conversion and remarketing pixels from firing. There is no conversion to credit back, and nothing for Smart Bidding to learn from.
• Lead Shield stops the form fill. Fake submissions never reach HubSpot or your sales team, so there is no junk lead to clean out of the CRM after the credit arrives.
• Coverage is not limited to Search and PMax. The same protection applies across Google, Meta and Microsoft, and WordPress Bot Mitigation covers organic and direct traffic at the site level.

What are advertisers and security teams saying?

The gap between “credited” and “clean” is not a vendor talking point. It shows up in Google’s own documentation, in practitioner case studies and in how marketing leaders describe the risk.

Google has put the limitation in writing. In September 2026 a Google Ads Help page titled “Prevent invalid leads” stated that Google’s invalid traffic filters don’t prevent invalid leads on their own, because most traffic checks happen after the ad serves. The page asks advertisers to add reCAPTCHA, double opt-in and server-side validation to their own forms, and frames lead quality as a shared responsibility. It is the same three-part advice the Google Ads Liaison gave in October 2022 to an advertiser complaining about junk leads from Performance Max. Four years on, the answer has moved from a social media reply to a standing help page, and the underlying fact has not changed: Google’s job ends at deciding whether a click is billable.

Marketing leaders rank the bidding problem above the budget problem. A July 2026 Lunio survey of 131 senior marketing leaders, reported by PPC Land, found that 51.1% named automated bidding optimising toward non-human converters as a top risk, ahead of sales time lost to bot-generated leads (22.9%) and corrupted CRM data (22.1%). In the same survey, 38.9% rated their trust in platform invalid click credits at 4 or below out of 10, yet only 5.3% ran a dedicated invalid traffic tool. In other words, most advertisers know the credit is not enough and are still relying on it.

Practitioners who push Google for a full investigation get a telling ratio. One small business documented its case in October 2026: over 60 days, Google classed 952 of 6,961 clicks as invalid and filtered them before billing, then credited a further 31 clicks after the fact. That is 12% caught at the door and 0.4% refunded later. The credit report is the 0.4%. Separately, a Spanish agency demonstrated in April 2026 that Google Ads IP exclusion lists can be bypassed by splitting the IP that harvests a click ID from the IP that later uses it, which is one reason agencies stopped treating native exclusions as a complete answer.

Security teams see the same traffic from the other side. Imperva’s 2026 Bad Bot Report put automated traffic at 53% of all web activity in 2025 and bad bots alone at 40%, and its prior report noted that bad bots particularly thrive on marketing campaigns, because paid campaigns are where the money and the forms are. A product CISO writing for MarketingProfs described routinely “plotting” with digital marketing leaders against bot traffic as a common enemy. For a CISO, a bot that clicks an ad and lands on a lead form is not a billing question; it is an unauthenticated client hitting a public endpoint, scraping content, probing for weaknesses and polluting the CRM. Google’s credit addresses none of that.

How should I use the report alongside ClickCease?

Pull it. It is free, it is Google’s own data, and it is the only official number you will ever get on how much invalid traffic got past Google’s front door. Then run one comparison every month:

1. Take credited amount from the Invalid Activity Credit Report for the last closed month, Search and PMax.
2. Take blocked spend (or blocked clicks × average CPC) from your ClickCease dashboard for the same month and the same campaigns.
3. The ClickCease figure is fraud that never cost you anything and never touched your data. The Google figure is fraud that cost you money for a month, was partly refunded, and is still in your reports, audiences and bidding history.

If you are not yet running protection, do the same exercise with the credit alone. Then ask what the credited clicks did to your conversion rate, your remarketing lists and your CPA target while they were live. The credit is the small number.

Frequently asked questions

What is the Google Ads Invalid Activity Credit Report?

It is a Report Editor template, documented by Google in June 2026, that breaks down credits for invalid activity by campaign and network for Search and Performance Max. It shows credited clicks, credited interactions and credited amount, plus “adjusted” versions of cost, clicks, CTR, CPC, conversion rate and cost per conversion.

Does Google refund all invalid clicks?

No. Google filters most invalid traffic in real time before billing and credits some of what it finds afterwards, “where appropriate and possible”. The report only covers that post-invoice portion. Clicks Google’s filters classify as valid are never credited, and investigations are limited to the last 60 days.

Does an invalid activity credit remove the click from my reports?

No. The original click stays in the month it occurred and the credit is applied in the following month. Google provides “adjusted” metrics in the report because standard campaign metrics are not corrected retroactively.

Does Smart Bidding unlearn clicks that Google credits back?

Google does not say. Its documentation describes the credit as a billing adjustment and publishes nothing stating that credited activity is removed from the signals Smart Bidding trains on. The safe assumption is that a click or conversion the model saw remains in its training history.

Doesn’t Google remove invalid clicks from Smart Bidding automatically?

Not that Google has documented. Google states that filtered invalid clicks are removed from campaign metrics, which is a reporting statement, and the same help text notes that in some cases the conversion from an invalid click remains. The tools Google offers for cleaning bidding data, data exclusions and conversion retractions, are manual and advertiser-operated, which is the clearest sign that no automatic purge exists for credited activity.

Which campaigns does the report cover?

Search and Performance Max only. Display, Video, Demand Gen and App campaigns are not included in the template.

If Google already credits invalid clicks, why do I need click fraud protection?

Because the credit only returns money for clicks Google detected after the fact. Prevention blocks invalid visitors before the ad serves or before the session registers, so they never consume budget, never appear in analytics, never fire conversion pixels, never submit forms, and never enter bidding data. A credit fixes the invoice; prevention fixes the account.

How do I compare Google’s credits with ClickCease’s blocked traffic?

Pull the Invalid Activity Credit Report for a closed month and note the credited amount for Search and PMax. In ClickCease, take blocked clicks or blocked spend for the same month and campaigns. The first is fraud Google refunded after it ran; the second is fraud that never ran.

Sources: Google Ads Help, About the Invalid Activity Credit Report, About invalid traffic, Invalid clicks: definition, About data exclusions and About conversion adjustments; PPC Land on the Invalid Activity Credit Report, the “Prevent invalid leads” help page, the Lunio July 2026 survey and the IP exclusion bypass; Imperva Bad Bot Report 2026; MarketingProfs, The Fight Against Bot Traffic; canadacreate.com case study, October 2026.