Why Google Ads Sends Bad or Spam Leads
Spam forms, fake details, wrong-service enquiries and invalid clicks are different problems. Learn how to identify which one is hurting lead quality.
By Gavin Sevastian · Updated August 11, 2026 · 14 min read
The phone rang. A form reached the inbox. A new contact appeared in the CRM. Then the supposed lead asked for a job, lived far outside your service area, wanted a service you do not offer, used someone else's phone number, or turned out to be a string of nonsense.
This page begins after an enquiry arrived. If your ads spend without producing enquiries at all, start with why your Google Ads are not getting calls. A quiet phone and a phone full of bad leads are different failures.
Spam is too broad to be useful. It can mean an automated form submission, invented contact details, a real person looking for the wrong service, a real person outside your service area, a low-intent shopper, an invalid ad click, or a conversion that was never a lead. Name the one you have before you try to fix it.
Ask what reached you, where it entered, and what the evidence rules in or out. That tells you whether to look at the form, campaign, page, location settings, measurement, or Google's traffic-quality team.
A bad lead is not one diagnosis
Start by separating who or what completed the action. Do not begin by deciding every bad outcome was a bot. Real people create plenty of useless enquiries, and valid ad clicks can still produce poor-fit leads.
| What arrived | What it usually means | Where the first useful evidence lives |
|---|---|---|
| Automated form submission | Software completed or replayed the form | Form logs, validation result, timing, and submission pattern |
| Invented identity | The details do not belong to a reachable prospect | Phone or email verification and the original form payload |
| Wrong-service person | A human misunderstood the offer or matched through an irrelevant search | Search term, ad message, and landing-page service promise |
| Out-of-area person | A human was outside the practical service territory | Customer address, campaign location settings, and location reporting |
| Low-intent shopper | A human is researching, comparing, or not ready | Call notes, form question, and the promise made before submission |
| Invalid ad interaction | The click or impression was not genuine interest | Google Ads invalid-traffic reporting and traffic evidence |
| Miscounted conversion | The dashboard recorded an action that was not a submitted lead | Conversion action, trigger, and the inbox or phone record |
The first five are lead-quality or customer-path problems. The sixth asks whether the ad interaction was valid. The seventh is measurement. A form filled by a bot may have followed an invalid click, a valid click, no ad click, or a visit from another channel. The submission alone cannot prove which one.
Bot and automated form spam starts at the form
Automated submissions often share obvious traits: impossible names, repeated text, unrelated pitches, or fields completed in unusual ways. That doesn't mean an ad click caused them. Public forms are discoverable by spam tools whether or not Google Ads is running.
Protection belongs on the server as well as the page
If your form only checks fields in the browser, it can still receive a direct automated request. Google's own invalid-traffic guidance recommends server-side validation for lead forms (Google Ads Help). That means the system receiving the submission checks that required data is present and plausible before treating it as a lead.
Risk tools can add another signal. reCAPTCHA v3 returns a score for an interaction and tells implementers to verify the result on the backend. Google also says every site is different and recommends observing real traffic before deciding how to act on scores. There is no universal number that separates every homeowner from every bot.
A score is evidence, not a verdict. Contractor sites receive rushed mobile visitors, people on shared networks, and privacy-conscious browsers. Rejecting all unusual behaviour can make spam look lower by throwing away valid enquiries too.
More friction reduces some spam and some real leads
Requiring more information can discourage automation and expose people who never read the offer. It can also cost a legitimate emergency lead who is standing in a flooded basement with one hand on a phone.
Your form needs to ask what you need to route the enquiry. Service and job location are useful because they change whether you can help. A long questionnaire added only to make submission harder is friction without information.
Verification has the same tradeoff. Confirming a phone number or email may reduce fake identities, but it inserts another step between intent and contact. Google currently offers lead form assets that let people submit information directly inside an ad, and describes OTP verification as one of their qualification features (Google Ads Help). That is a current product feature, not a guarantee that every verified lead will be a good job.
Fake details need confirmation before they need a campaign rebuild
There are two common versions of a fake lead. In one, the contact details are invented and nobody can be reached. In the other, the phone number or email belongs to a real person who says they never asked for service. The second deserves care because a real identity may have been entered by someone else.
Preserve the source before deciding what happened
Keep the original submission time, landing page, source information, and submitted fields long enough to compare them. If the contact arrived from a Google-hosted lead form asset, its path differs from a submission on your website. If it came through a website form, it may have arrived from paid search, organic search, a referral, or direct automated traffic.
That is why "It showed up in our leads inbox" is not enough attribution. The inbox proves delivery, not source. Before changing keywords or requesting an invalid-traffic review, establish whether the lead can actually be connected to an ad interaction.
Do not treat repeated IP addresses as proof of fraud. Google notes that return visits, shared connections, rotating addresses, and third-party reporting can make multiple visits appear under one IP (Google Ads Help). An IP can support an investigation, not decide it.
Real people can still be the wrong leads
A human can complete every field honestly and still be a poor lead. This is the largest category hidden by the word spam because it feels easier to blame automation than to admit the campaign invited the wrong person.
Wrong-service leads usually begin before the form
Compare the words the person searched with the service named in the ad and the service explained on the page. A broad service label can mean different things to a homeowner and a contractor. Restoration can mean water removal, reconstruction, contents cleaning, mould work, or fire damage. General contractor can mean a residential renovation, a commercial tender, or someone looking for employment.
Narrow the promise in your ad before the click. Repeat the included service on the landing page and make important exclusions visible before the form. If the campaign advertises work you do not perform, adding form protection treats the wrong layer.
Negative keywords can reduce irrelevant searches, but a complete public list would be brittle and dangerous. The same word can describe wanted and unwanted work in different queries. Search term, match, ad group, and offer need to be read together.
Low intent is not the same as fake intent
Homeowners compare companies, ask for rough prices, submit after hours, and change their mind by morning. Some are early in a long project. Those leads may be frustrating without being fraudulent or irrelevant.
The page can qualify timing, scope, location, and service without demanding that every visitor be ready to sign. If a sale normally takes several conversations, judging every first enquiry as poor because it did not book immediately misreads the cycle.
This distinction is where a focused landing page earns its keep. It can answer one service decision clearly without forcing a paid visitor to navigate a general company site.
Wrong-location leads have several possible causes
An out-of-area enquiry can come from broad location settings, imperfect location signals, a search that names the target area, an inaccurate website promise, or a person arranging work at a property somewhere other than their current location.
Location targeting reduces mismatch but cannot guarantee location
Google's default broad option can include people who are in, regularly in, or interested in a targeted place. The narrower presence option limits targeting to people likely to be in or regularly in the area (Google Ads Help). For a contractor who travels to the property, that choice can materially change fit.
Google also says it uses several signals to infer location and that accuracy is not guaranteed in every situation. Its separate guidance on minimizing clicks outside targeted areas says presence targeting can help, not that every outside-area click will disappear.
The customer address remains the operational truth. Someone searching from outside the area may own a local property and be a valid lead. Someone physically inside the target may want work at a distant property. Geography should be judged against the job address and service boundary, not only the device location in a report.
Language mismatch can look like poor lead quality
Your campaign may reach someone who can read the ad but cannot comfortably complete the sales conversation in the language you support. The landing page, form, confirmation, and phone response need to make the same language promise.
Adding translated keywords and ads without a matching page and follow-up path creates enquiries your office cannot handle. That is not bot traffic. It is a broken promise between the ad and the response.
Separate the lead source and network before judging either
Not every Google Ads lead follows the same path. Search text ads can appear on Google search properties and on Google search partner sites. Google currently describes search partners as hundreds of non-Google websites plus Google Video and other Google sites (Google Ads Help).
Your website forms, calls from ads, landing-page calls, and Google-hosted lead forms carry different context and friction. A lead form asset can submit without a site visit. A website form can collect details the ad form did not. Neither is automatically better.
Compare source quality without declaring an entire source fraudulent
Break the evidence out by network, campaign, form path, service, and geography. Look for where poor outcomes cluster. Then compare qualified and booked outcomes, not submission count alone.
If one of your sources produces weaker leads, inspect the promise and path before assuming fraud. A lower-friction form can produce more early-stage enquiries. A broader network can reach a different context. A mislabeled conversion can make one source appear to produce leads it never delivered.
Make your network and asset decisions from real business outcomes. Product behaviour changes, so current Google documentation and the live account beat an old tutorial.
Google needs a better outcome than form submitted
If you send every submission back to Google as equally successful, automated bidding has no way to distinguish a booked replacement from spam, a wrong-service request, or a person who never answers.
Qualified and converted leads provide a stronger signal
Google's enhanced conversions for leads can connect offline outcomes back to advertising using user-provided data, and Google recommends distinct qualified-lead and converted-lead goals for lead-generation advertisers (Google Ads Help). The concept matters more here than the implementation: tell the account which enquiries progressed instead of rewarding every form equally.
You need stable definitions, appropriate consent and privacy handling, and a reliable connection between the original enquiry and later outcome. The technical method changes, so a permanent step-by-step recipe would age badly.
Better feedback cannot repair a misleading ad or broken form. It gives the bidding system a better target after the customer path is sound.
Invalid traffic is narrower than bad lead quality
Google defines invalid traffic as clicks and impressions that are not the result of genuine user interest, including accidental or duplicate clicks, fraudulent clicks, and automated tools such as bots or crawlers (Google Ads Help).
Google says its systems evaluate interactions continuously and filter invalid activity. Activity found before invoicing is adjusted before billing. Activity identified after an invoice can appear as an invalid-activity credit on a later invoice. Google explicitly describes these as adjustments or credits, not refunds.
A useless lead can follow a valid click
A real homeowner can click your ad, misunderstand the service, and submit a useless request. A comparison shopper can click several ads. Someone outside the service area can search for work at a local property. None of those facts alone makes the click invalid.
The reverse can also happen: an interaction may be filtered as invalid while a later conversion record remains. Google notes this rare possibility in its invalid-traffic guidance. That is another reason not to force click totals and lead records into a simple fraud story.
An investigation needs evidence, not certainty from one signal
Google provides an Invalid clicks column, billing reports for invalid-activity credits, and a Click Quality Form for suspected invalid clicks, poor-quality form submissions, outside-location clicks, or sudden cost changes (Google Ads Help). Its invalid-traffic guidance says manual investigations cover traffic from the previous 60 days.
When you request an investigation, identify dates, campaigns, sources, and evidence connecting the suspicious activity to your ads. The request does not promise that Google will classify the activity as invalid or issue a credit. Repeated IPs, unusual timing, and form patterns can support it, but each has innocent explanations on its own.
Route the symptom to the layer that can fix it
Use this as a one-time map, not a weekly ritual. Stop at the first place where the evidence clearly breaks.
Confirm that a lead really arrived
Match the dashboard conversion to the phone, form, inbox, or CRM. If there is no enquiry, the problem is measurement rather than lead quality.
Identify the submission path
Separate website forms, Google-hosted forms, calls, and other sources. An inbox record alone does not establish paid-search attribution.
Classify person, fit, and location
Was it automated, an invented identity, a real wrong-service person, an out-of-area job, or an early-stage shopper? Keep those labels separate.
Check the promise before the protection
Compare search, ad, page, and form. A wrong-service human needs a clearer ad and page; an automated submission needs form-side controls.
Compare source and network outcomes
Look for a pattern by campaign, network, geography, and lead path without declaring fraud from one weak cluster.
Escalate suspicious traffic only with connected evidence
Use Google's traffic reports and investigation channel when the suspicious interaction can be tied to the ad account.
The order prevents rebuilding a campaign because a form is unprotected, or adding form friction because the ads invite the wrong service. It also separates conversion counting from lead quality before bidding learns from the wrong result. Common Google Ads mistakes contractors make covers the wider account failures.
Questions contractors ask about Google Ads spam leads
Common questions
First establish whether the submission can be tied to an ad. Public website forms receive automated traffic from many sources. If it is connected, the cause may be invalid traffic, a low-friction lead form, broad matching, weak qualification, geography, or a conversion that counts the wrong action. Those require different evidence and fixes.
No. A form submission does not prove how your lead arrived or whether Google classified an ad interaction as valid. Match the submission to source and click evidence. Google automatically filters activity it determines is invalid, while a valid click can still produce a useless submission.
Google says invalid activity is handled through billing adjustments or credits, not refunds. Activity detected before invoicing is adjusted before billing; activity identified later may appear as a credit. There is no promise that a suspicious lead will be classified as invalid.
No. It adds a risk signal and requires correct backend verification. Google says every site's traffic is different, so there is no universal score that safely blocks every bot without losing real people. Your form protection, validation, and lead-source evidence still matter.
Add questions that change routing or qualification, such as service and job location. Extra friction can reduce automated or casual submissions, but it can also lose valid mobile and emergency leads. Make every required field earn its place.
Possible causes include the broad presence-or-interest location option, imperfect location signals, a person searching for another property, or a website promise wider than the practical service area. Judge the job address separately from the searcher's inferred location.
They can produce a different lead mix, but the source is not proof of fraud. Compare your qualified and booked outcomes by network and campaign, then inspect the ad and lead path. Make a network decision from observed outcomes, not a blanket assumption.
Google supports qualified-lead and converted-lead outcomes through enhanced conversions for leads. You still need consistent definitions, reliable outcome capture, and appropriate privacy handling. Better feedback helps only after the campaign and customer path are sound.
Better lead quality starts with naming the failure correctly
Bot spam belongs at the form and server. Invented identities need source confirmation. Wrong-service humans point back to search, ad, and page clarity. Wrong-location enquiries need a location check. Early shoppers need a sales-cycle judgment. Miscounted conversions belong in tracking. Suspected invalid interactions belong in Google's traffic-quality path.
Calling all seven spam feels decisive and fixes none of them. The useful work is matching each enquiry to the layer that produced it, then changing only that layer.
If your Google Ads account produces activity but the leads do not resemble the work you want, start with a free account review. I will trace the path from click to enquiry and tell you where the mismatch begins, without pretending every bad lead is fraud.
