Google Ads click fraud for contractors: what is actually invalid?

A sudden run of expensive clicks with no booked work deserves investigation, but it does not prove click fraud. Google defines invalid traffic as clicks and impressions that do not result from genuine user interest, including deliberate fraud and accidental or duplicate activity. Contractors should first reconcile Google’s invalid-traffic data, queries, networks, locations and real lead outcomes before buying a blocking tool or accusing a competitor.

Click-fraud suspicion is understandable in home services. A single emergency roofing, HVAC or plumbing click can be expensive, competitors are visible in the same local auctions, and a quiet phone makes every charge feel suspect. The useful question is not “Do bots exist?” They do. The useful question is “Which activity was invalid, what did Google already remove, and what controllable waste remains?”

What counts as invalid traffic?

Google’s definition is narrower than “a click that did not become a lead.” It includes intentional fraudulent traffic, accidental clicks and duplicate clicks that do not represent genuine interest. Examples Google gives include competitors clicking to raise costs, publishers encouraging clicks and automated traffic.

A real homeowner can click and still be a poor prospect. Someone may want a service the contractor does not offer, live beyond the dispatch area, be comparing prices, need help next year or abandon a slow page. Those visits may waste budget, but they are not automatically invalid under the platform definition.

What does Google filter and credit?

Google says its systems evaluate interactions using multiple layers of automated filters, machine learning and manual review. Traffic detected as invalid in real time is filtered before the advertiser is charged. When qualifying invalid activity is detected later, Google says a credit may be issued in a subsequent billing cycle where appropriate and possible.

That means a contractor should not calculate suspected loss by multiplying every “Invalid click” by average cost per click. The invalid-click column represents activity Google says it has already identified and filtered. Late adjustments can appear separately in billing, while the Invalid Activity Credit report can show campaign metrics adjusted for credited traffic.

Ask an agency to show all three, for the same date range:

• Invalid clicks and invalid click rate: add the columns at campaign level.
• Invalid Activity Credit report: review later-classified traffic and affected campaigns.
• Billing adjustments: reconcile credits rather than assuming every suspicious visit was charged.

Google notes that third-party monitoring tools may report different numbers because their methods and definitions differ. A discrepancy is a reason to examine evidence, not proof that either number represents recoverable spend.

Bad traffic is not always click fraud

For contractors, ordinary campaign mistakes often produce the same symptoms as fraud. Broad matching can reach employment, training, parts, DIY and research searches. Location settings can reach people interested in a target area rather than physically present there. Search Partners can introduce traffic from non-Google properties. A primary conversion set to page views or button clicks can make automated bidding chase activity that never becomes work.

Those problems are controllable. Review the search-term and negative-keyword audit, check location targeting against actual job addresses, and compare Google Search with Search Partner outcomes. None of those actions claims to block fraud; they remove avoidable ambiguity.

Lead handling belongs in the same diagnosis. A genuine emergency caller who hangs up after three unanswered rings is not an invalid click. Neither is a form that sat in an inbox until the next morning. Match every click investigation to connected calls, qualified enquiries, bookings and wins before concluding the traffic itself was fake.

A practical seven-part invalid-traffic audit

1. Define the event. Record the affected campaigns, dates, budget change, click spike and business symptom. “Leads feel bad” is too broad.
2. Verify charges and credits. Export clicks, invalid clicks, invalid rate, costs and the Invalid Activity Credit report. Check billing adjustments.
3. Segment the traffic. Compare campaign, network, device, hour, day, location and search term. Look for concentration without treating correlation as proof.
4. Reconcile site evidence. Review analytics and server or form logs while allowing for consent loss, tracking failures and calls that happen without long sessions.
5. Classify every lead. Use spam, applicant, supplier, existing customer, wrong service, wrong area, qualified, booked and won—not simply “bad.”
6. Remove ordinary waste. Fix queries, locations, networks, forms, page performance and phone coverage first. Preserve a change log.
7. Escalate a documented anomaly. If unusual activity remains, give Google Ads support dates, campaigns, click IDs where available, IP or log evidence and a clear description rather than a screenshot of a high bounce rate.

This sequence protects the contractor from two expensive errors: ignoring a genuine traffic-quality problem and paying for a fraud product when the actual leak is basic account management.

Are IP exclusions useful?

Google Ads allows eligible campaigns to exclude specific IP addresses. Google currently documents up to 500 addresses per campaign and also supports account-level exclusions. The feature is not available for every campaign type, including Performance Max.

Use it narrowly. A known office network, agency network or stable address associated with verified unwanted activity can be a reasonable exclusion. Mobile carriers, shared Wi-Fi, dynamic home connections, VPNs and corporate networks can put many legitimate people behind changing or common addresses. Blocking broad ranges on weak evidence can remove customers while creating a false sense of protection.

An IP address also does not tell you intent. Repetition from one address may be a property manager comparing options, a homeowner returning across several days or an entire workplace sharing a network. Pair network evidence with time, query, campaign, device and downstream behavior.

When should a contractor consider third-party software?

A specialist monitoring tool can be useful when spend is material, anomalies repeat, the business has enough data to establish a baseline and the vendor explains what it detects and what it changes. It should add evidence and controlled response—not replace Google Ads fundamentals.

Before signing, ask:

• Does the tool monitor Search, Search Partners, Display and Performance Max, or only some traffic?
• What exactly qualifies as invalid, suspicious or blocked?
• How does it control false positives on mobile, shared and dynamic IP addresses?
• Does it alter exclusions or campaign settings automatically?
• Can you inspect event-level evidence rather than a headline “fraud rate”?
• How are consent, privacy and data retention handled in the US, Canada and UK?
• Can results be tested against qualified leads and booked jobs?

Do not accept guaranteed savings, guaranteed refunds or an industry-wide fraud percentage as proof for your account. The relevant baseline is the contractor’s own campaigns and outcomes.

What should an agency report?

A useful monthly report separates four layers: Google-filtered invalid activity, later credits, suspicious traffic under investigation and valid but unprofitable demand. Combining them into one “fraud” number prevents sensible action.

• Invalid clicks and invalid click rate by campaign
• Invalid Activity Credit report and billing reconciliation
• Network, device, location and time concentrations
• Search terms and landing pages associated with the anomaly
• Spam, wrong-area, wrong-service, qualified, booked and won outcomes
• Changes made, evidence retained and the next review date

Owners should also ask whether the agency has account access and change-history controls. A sudden performance shift after a network, matching, bidding or conversion change needs a different remedy from automated abuse.

US, Canada and UK considerations

The audit method is the same across all three markets, but operational boundaries differ. US campaigns may span states, licensing areas and large metros. Canadian accounts must account for provinces, long dispatch distances and bilingual demand. UK contractors often diagnose by towns, counties and postcodes rather than broad radii.

Privacy and consent obligations also differ and can affect analytics completeness, log retention and the use of third-party monitoring. Treat legal compliance as a local implementation question. This article recommends measurement discipline; it does not prescribe a universal tracking or retention setup.

Frequently asked questions