Monday, 7:15 a.m., at a plumbing company in Katy. The owner opens the CRM to 23 weekend form fills from Google Ads. Nine are gibberish (“asdf asdf,” a phone number of 1234567890). Four are offers to “rank your website #1.” Three are real people who wanted a water heater quote and never got a call back because everyone assumed the whole batch was junk. The rest are hard to tell.
By 7:30 he’s convinced the guy across FM 1463 is clicking his ads.
Maybe. More often, what’s going on is a mix of three separate problems that get lumped together as “click fraud”: invalid clicks that Google mostly catches and doesn’t bill, form and call spam that Google can’t see at all, and campaign settings that buy cheap, low-quality traffic on purpose. Each has a different fix. The worst move is the common one, which is counting all those junk form fills as conversions and letting Smart Bidding go find more of them.
Below: what Google actually filters, where the junk comes from, how to block it, and how to file a claim when you think Google missed something.
Invalid clicks, spam leads and bad targeting are three different problems
Google defines invalid traffic as traffic that “doesn’t represent genuine interest in your business,” and its examples include bots, accidental clicks, fraudulent placements like clickjacking and ad stacking, and deliberate clicks by competitors or people paid to click, per Google’s invalid traffic help page. That’s the click side.
The lead side is different. A bot that skips your ad entirely and hammers your contact form isn’t a click problem, and neither is a human in another country filling out forms for a few cents each. Google says plainly that it can’t control what visitors do after the click, and it recommends reCAPTCHA, server-side validation and double opt-in for invalid leads (same page).
Then there’s the third bucket, which isn’t fraud at all: real people who were never going to buy. Somebody who searched “plumber salary Houston” and clicked, or a kid on a mobile game who tapped a Display ad by accident. Those are targeting problems, and they’re usually the biggest of the three.
Sorting your junk into these buckets is the first job, because the fixes don’t overlap much. A click-fraud tool won’t stop form spam from a bot that never clicked an ad. A reCAPTCHA won’t stop you from paying for clicks on a parked domain.
What Google already filters, and how to see it in your account
Google runs automated filters and machine learning on every click, and its Ad Traffic Quality team does manual reviews on top of that. Clicks it flags before your billing cycle closes are never charged. Clicks it catches later are credited “when appropriate and possible,” usually in the next billing cycle, and they show up on your invoice as “Invalid activity” adjustments, per Google’s invalid traffic help page.
You can see the filtered clicks yourself. Most owners never have.
- In Google Ads, go to Campaigns in the Campaigns menu.
- Click the Columns icon above the table.
- Search “Invalid clicks” in the column picker and add Invalid clicks and Invalid click rate.
- Click Apply, then set the date range to the last 90 days.
Those numbers are clicks Google already threw out. You weren’t charged for them, and they aren’t in your Clicks column. Google’s Overview page is also rolling out an Invalid traffic card with “Estimated savings” and “Credited cost,” and the credits themselves live under Billing > Summary, in the Adjustments dropdown on the Month card.
A couple of things Google’s page says that are worth knowing before you panic. There’s no industry standard for a “normal” invalid click rate. Traffic spikes aren’t automatically fraud; demand, competitor budget changes and your own keyword changes cause them too. That matters in Houston, where a freeze warning or a storm in the Gulf can triple plumbing and roofing searches overnight. And zero-second visits in your analytics are often real people whose session never registered a second hit, while shared IP addresses are normal on office and mobile networks that route lots of users through one address.
Is a competitor clicking your ads?
Sometimes, yes. Google lists deliberate competitor clicks as invalid traffic, so it isn’t a myth. In local accounts we’ve run, though, the competitor clicking an owner is sure about rarely shows up in the data. What does show up is the competitor’s office manager searching your company name once a week to see your ads, which costs you a few dollars a month. Repeated clicks from the same source on the same day are exactly the pattern Google’s filters are built to catch.
If you suspect someone specific, the evidence you want is in your web server logs or analytics: the same IP address or device hitting your landing page from paid clicks several times in a short window, with no form fill and no call.
IP exclusions have real limits, and they’re worth knowing before anyone sells you a tool built around them. Per Google’s IP exclusion help page:
- You can exclude up to 500 IP addresses per campaign. An asterisk in place of the last part of an address blocks a range.
- An account-level IP exclusion list (Admin > Account settings > IP exclusions) applies to every campaign, including Performance Max, Demand Gen, Search, Shopping, Display and YouTube.
- Campaign-level exclusions (campaign Settings > Additional settings > IP exclusions) aren’t available for video, hotel, app, Performance Max or Smart Display campaigns.
- If both lists exist, Google merges them.
Now the practical problem. Most phone traffic in Houston comes through carrier networks where thousands of people share and rotate addresses, so blocking a mobile IP can block your neighbors and won’t stop the person you were after for long. IP blocking works well against a fixed office connection or a hosting provider’s range. Against a determined person on a phone, it barely works. And don’t block your own office: just have your staff stop clicking your ads to “check on them.” Use the Ad Preview and Diagnosis tool instead.
Search Partners, Display and Performance Max: where most junk comes from
When we look at a lead-gen account full of spam, the source is usually network settings rather than a fraud ring. A Search campaign created with the defaults opts you into Google Search Partners and often the Display Network too. Performance Max serves across Search, Display, YouTube, Gmail, Maps and partner sites by design.
| Source | What the junk looks like | The control |
|---|---|---|
| Search Partners (Search campaigns) | Cheap clicks, low engagement, odd form fills at odd hours | Segment by Network; uncheck “Include Google search partners” in campaign Settings > Networks if it doesn’t produce real leads |
| Display Network (opted in from a Search campaign) | Accidental taps in mobile games and apps, near-zero time on site | Uncheck “Include Google Display Network” on Search campaigns; run Display separately, on purpose, if at all |
| Performance Max placements | Lead forms from apps and low-quality sites, spam that converts “cheaply” | Account-level placement exclusions, brand-safe content settings, offline conversion data, and the newer channel controls |
| Google lead form assets | Fast, prefilled submissions from people who barely read the ad | Add qualifying questions and call every lead fast; judge on booked jobs |
| Your own website form | Bot submissions that never came from an ad at all | reCAPTCHA, honeypot, server-side validation (next section) |
Google has added more visibility here. In August 2025 it announced full placement reporting for the Search Partner Network across Search, Shopping and App campaigns, listing each partner site with impressions, according to PPC Land. It ended parked domains as a Search Partner surface on February 10, 2026, and in October 2026 it started rolling out channel prioritization controls in Performance Max that let you set Search partners and Display to low priority with a best-effort budget cap, per PPC Land’s timeline of the Performance Max partner controls.
For placements you’ve identified as junk, account-level placement exclusions are the tool. Per Google’s account-level placement exclusion page, they live under Tools > Content suitability > Advanced settings > Excluded placements, they override campaign settings, they apply to Performance Max, they’ve covered Search Partner sites since March 2024, and you can have up to 65,000 per account.
Our rule of thumb for a local service business: run Search with partners off until you have clean conversion data, then test partners on in one campaign and compare cost per booked job, which is a different number from cost per form fill. We cover the network checks in more detail in our clicks-but-no-calls audit, and the Search vs. PMax trade-off in Performance Max vs. Search for local businesses.
Stop form spam at the form
Form spam is the one problem you own outright. Google can’t filter a bot that submits your contact form directly, and plenty of spam on a typical Woodlands or Sugar Land service-business site never touched an ad. It still pollutes your lead count, and if your conversion tag fires on the thank-you page, it pollutes your Google Ads data too.
The fixes stack. None of them is enough alone.
A few notes on the list. Google’s reCAPTCHA v3 documentation describes v3 as invisible to visitors and returning a score where 1.0 is very likely a good interaction and 0.0 is very likely a bot; Google suggests starting your cutoff at 0.5 and adjusting from your own traffic. Google has since moved reCAPTCHA documentation under Google Cloud, so check which version your form plugin uses. A honeypot is a hidden field that people never see and bots fill in; any submission with that field filled gets dropped. Server-side validation means your website rejects a fake phone number or a disposable email domain before the lead is stored, not just with a JavaScript warning a bot can skip.
The biggest single change, though, is moving the conversion. If Google Ads counts a conversion every time the thank-you page loads, every bot that reaches it is a “lead.” Fire the conversion only on validated submissions, and then send back which leads were real (the offline conversion section below). Our form and call tracking guide shows the setup in GA4 and Tag Manager.
One caution: don’t add so much friction that real customers give up. A homeowner in Spring with water on the kitchen floor won’t solve a puzzle. Invisible checks first, and a visible challenge only for submissions that already look suspicious.
Call spam and junk calls
Calls are harder to fake at scale than forms, but Houston businesses still get plenty of junk: robocalls offering business loans, “Google listing” scam calls, wrong numbers, and real people looking for a different company with a similar name.
For years, Google Ads counted a call as a conversion if it lasted longer than a minimum duration you set, per Google’s call tracking help page. That meant a three-minute robocall pitch could count as a lead. In April 2026, Google documented AI-qualified call conversions: when call recording is on, Google reviews the recording for signs of real interest and only counts calls that pass, falling back to the duration threshold when a call can’t be recorded, per PPC Land. Recording only works when both numbers are in the U.S. or Canada, and the setting is under Admin > Account settings > Call ads > Call recording.
What we’d do for a Houston service business:
- Check your call conversion action’s duration threshold. If your real booking calls take four minutes, a 30-second threshold is counting a lot of noise.
- Use a call tracking platform with recordings and tag each call (booked, quote, spam, existing customer, wrong number). Our call tracking guide covers options.
- Schedule call assets only for hours someone answers, so after-hours spam doesn’t ring to a voicemail nobody checks.
- Don’t confuse ad-driven spam with Business Profile scam calls, which have their own fix in our post on Google listing scam calls.
Train bidding on real leads with offline conversion imports
This is the fix that matters most and gets skipped most. Smart Bidding strategies like Maximize conversions and Target CPA chase whatever you tell them is a conversion. If spam form fills count, the algorithm learns that the cheapest “conversions” come from junk placements, and it buys more of them. You end up with a lower cost per lead and fewer customers. We see this constantly in accounts we take over.
The answer is to tell Google which leads turned into something real. Two ways to do it:
- GCLID import. Your form captures the Google click ID, your CRM stores it, and you upload qualified leads and closed jobs back to Google Ads. Per Google’s offline conversion import FAQ, Google keeps the click ID for 90 days, so uploads must happen within 90 days of the click.
- Enhanced conversions for leads. Instead of the click ID, Google matches hashed email or phone numbers from your form to the click. Google’s FAQ recommends starting here if you haven’t set up offline imports yet, because it’s easier, and notes uploads with that personal data can go back 63 days.
Google also recommends uploading daily for Smart Bidding, or on a steady schedule if you can’t, and uploading a new conversion action consistently for one to two conversion cycles before you make it the one bids optimize to (same FAQ).
Once real lead and booked-job conversions are flowing, change your primary conversion goal to the qualified one and demote raw form fills to secondary. It’s the same idea we describe for Meta in how to fix bad Facebook lead quality: the platform finds more of what you reward. Fast follow-up matters here too, since a lead nobody calls can’t be qualified. See our speed-to-lead post.
Third-party click-fraud tools: when they’re worth paying for
There’s a whole category of click-protection software (ClickCease, ClickGUARD, Fraud Blocker, Lunio and TrafficGuard are names you’ll run into). We don’t have affiliate deals with any of them, and we don’t think every account needs one.
What they generally do: put a script on your site, fingerprint visitors who arrive from paid clicks, score them, and push suspicious IP addresses into your Google Ads exclusion list automatically. Several also help you assemble refund claims.
The honest caveats:
- They’re working with the same 500-per-campaign IP limit and the same rotating mobile addresses described above.
- Their dashboards often count clicks Google already filtered and didn’t charge you for, which makes the “fraud saved” number look bigger than the money you’d actually have lost.
- They don’t fix bad network settings or thank-you-page conversion tags, which cause more junk than fraud does in most local accounts.
Where they can pay for themselves: high-CPC niches like legal, emergency restoration or HVAC in July, where a few dozen bad clicks a month cost real money, or accounts that keep seeing the same suspicious patterns after the basic fixes are in. If you try one, use the trial period to compare your cost per booked job before and after. If it doesn’t move that number, cancel it.
How to request an invalid activity investigation
If you’ve done the basics and still believe Google charged you for clicks it shouldn’t have, file a claim. Per Google’s invalid traffic help page, you submit it through Google’s Click Quality Form, it covers activity from the last 60 days, reviews typically take several business days, and the findings come by email.
Gather this before you open the form, from that same Google page: your 10-digit Customer ID, the exact date range, the campaign and ad group names, the keywords you’re suspicious of, web server logs showing IP addresses, devices and GCLIDs, any relevant lead details, and the trends that made you suspicious. A vague “I think my competitor is clicking” claim with no logs rarely gets anywhere. A spreadsheet of 40 paid clicks from three IPs in two hours, with GCLIDs, is something an investigator can check.
Set expectations. Google says investigations won’t confirm whether specific clicks were marked valid or invalid, and credited clicks don’t retroactively change your campaign metrics, which are frozen at month end. If they find something, the credit appears as an Invalid activity adjustment. And since the 60-day window is fixed, don’t sit on a suspicion for a quarter.
A first-month cleanup plan
If you’re looking at a messy account right now, this is the order we’d go in.
Week 1. Add the Invalid clicks columns and check the last 90 days. Segment every Search campaign by Network and look at Search Partners and Display separately. Pull a list of the last 100 leads and tag each one real, spam or unclear. That gives you a baseline.
Week 2. Turn off the Display Network on Search campaigns. Turn off Search Partners on any campaign where partner leads are mostly junk. Add account-level placement exclusions for the worst Performance Max placements. Add reCAPTCHA v3 and a honeypot to every form, and move the conversion tag so it fires only on validated submissions.
Week 3. Set up enhanced conversions for leads or GCLID capture. Start tagging leads as qualified in your CRM and uploading them daily. Fix call conversion thresholds and turn on call recording if your business type allows it.
Week 4. If anything still looks wrong, pull the server logs and file a Click Quality Form claim inside the 60-day window. Then decide whether a third-party tool is worth a trial.
After that, judge everything on cost per booked job. If the spam is gone and leads still aren’t closing, the problem has moved to the landing page or the phones, and our Quality Score guide is a good next read. Our search marketing team does this cleanup for businesses across Greater Houston; if you’d like a senior strategist to look at your account and build a plan around your goals, get in touch.
Frequently asked questions
Does Google refund invalid clicks?
Google doesn’t charge for clicks its filters flag before your billing cycle closes, so no refund is needed for those. Clicks found invalid later are credited when appropriate and possible, usually in the next billing cycle, and appear as Invalid activity adjustments under Billing, Summary. You can see filtered clicks by adding the Invalid clicks and Invalid click rate columns at the campaign level.
How do I know if a competitor is clicking my Google Ads?
Look for repeated paid clicks from the same IP address or device in a short window, with no calls or form fills, in your server logs or analytics. Google treats deliberate competitor clicks as invalid traffic and filters many automatically. If the pattern comes from a fixed office connection, exclude that IP. If you believe you were charged, file a Click Quality Form claim within 60 days.
How many IP addresses can I exclude in Google Ads?
Google allows up to 500 IP address exclusions per campaign, and you can block a range by using an asterisk for the last part of an address. An account-level IP exclusion list applies to every campaign type, including Performance Max. Campaign-level exclusions aren’t available for video, hotel, app, Performance Max or Smart Display campaigns. Mobile IPs rotate, so blocking them is unreliable.
Why am I getting so many spam leads from Google Ads?
The usual causes are Search Partners and Display placements opted in by default, Performance Max serving on low-quality apps and sites, and bots submitting your website form directly. Counting every form fill as a conversion makes it worse, because Smart Bidding learns to find more cheap junk. Add reCAPTCHA and a honeypot, tighten network settings, and import qualified leads as your main conversion.
Are click fraud protection tools worth it?
Sometimes. They fingerprint paid visitors and add suspicious IPs to your exclusion list automatically, which helps in high-cost niches like legal, restoration or summer HVAC. They still face the 500-IP limit and rotating mobile addresses, and they don’t fix network settings or bad conversion tracking. Use a trial and compare cost per booked job before and after. Keep it only if that number improves.
