The symptom is familiar to anyone who runs lead generation. The Conversions column says one number, and sales can find far fewer leads in the CRM for the same week. Most people assume bots. Often the cause is the account itself, counting the same lead twice or counting things that were never leads.
This post covers the conversion-action side: where fake conversions come from, how to find them, which settings stop them and how to clean up what already counted. For how fake leads reach your forms and teach bidding the wrong customer, start with our guide to fake leads in Google Ads. The short version of that loop: whatever counts as a primary conversion becomes the customer the bidder looks for, so a junk conversion is a bad instruction.
What causes Google Ads fake conversions?
The five common sources of fake conversions are duplicate tags, test traffic, bot form fills, the wrong counting setting and imported junk. Only one of them is fraud. The other four are setup, which is good news, because setup is yours to fix.
Duplicate tags
Duplicate tags happen when two tags fire on one event, so one lead counts twice; most Google Ads duplicate conversions start here. The common version is a native Google Ads conversion tag on the thank-you page plus a GA4 key event for the same form, imported into Google Ads, with both set as primary in the goal you bid on. One lead, two conversions. Thank-you page reloads add more if the action counts Every. A transaction ID helps: Google says that "If there are 2 conversions for the same conversion action with the same transaction ID", it treats the second as a duplicate. The same page warns that adding one "will not resolve duplicate view-through conversions" (Use a transaction ID to minimize duplicate conversions, Google Ads Help).
Test traffic
QA submissions, staff demos and agency checks all fill real forms. Google's counting page says conversions count when they happen "after an ad interaction". In our reading, not Google's statement, a Google Ads test conversion only counts if the tester's browser clicked or interacted with one of your ads, for example someone clicking your own ad to check it and then submitting the form. We found no Google Ads feature that excludes internal or test conversions, so prevention sits in your tag rules and your test process. IP exclusions stop your ads showing to an IP address. They do not stop a conversion tag firing. GA4's internal-traffic filter drops matching data before Analytics processes it (Filter out internal traffic, Analytics Help), so in our understanding it only affects conversions you import from GA4, not native Google Ads tags.
Bot form fills
Some junk is real fraud. Google's lead-quality page says that "users or bots can still reach your website and submit fake or low-quality lead forms" (Prevent invalid leads, Google Ads Help). These submissions arrive through paid clicks and fire your conversion tag like any lead. They are one form of invalid traffic, and the form-side defences are covered in the bot form submissions section of our fake leads guide, linked above.
The wrong counting setting
A lead action set to Every can record several conversions for one person. A micro-conversion set as primary, such as a page view, a scroll or a tap on a click-to-call button, counts behaviour that is not a customer action at all. Both inflate the column without a single fake person involved. The fix is in the section on primary actions and counting below.
Imported junk
Offline or CRM imports can import every lead as a conversion, including unqualified and spam records. GA4 soft events imported as primary do the same. If you import CRM stages, our guide to enhanced conversions for leads covers which stages to send.
How do you find fake conversions in your account?
Open the conversion summary, compare each primary action with what really happened, then segment campaigns by conversion action. Google's summary page says: "Go to Conversions within the Goals menu" (Manage conversion goals from the conversions summary, Google Ads Help). Its primary and secondary page puts the summary in the same place, "housed in the Conversions drop-down in the Goals menu". So the path is Goals, then Conversions, then Summary. Next, segment the Campaigns table: select the segment icon, then Conversions, then Conversion action, to see which action drives the count (Use segments in your tables, Google Ads Help). Conversion segments only fill conversion columns, so clicks and cost show no data in that view.
Start with the two columns. Google reports primary actions in the Conversions column and secondary actions in the All conversions column (About primary and secondary conversion actions, Google Ads Help). The gap between them is what you observe but do not bid on. Then compare Conversions with your own form submissions or CRM leads for the same dates. If Conversions is higher, something primary is counting extra.
Check the status of each action too. The summary page lists "No recent conversions", "Inactive" and "Needs Attention" as signs of a problem, and says the last two usually point to tag misconfiguration, a missing tag or an enhanced conversions issue. A tag that broke and was fixed is a candidate for the data exclusion section further down.
One oddity is not a fake. Google notes that rarely your reports might show more conversions than clicks, because Google removes the invalid click from campaign metrics while the valid conversion remains (Invalid clicks: Definition, Google Ads Help). More conversions than clicks alone does not prove duplicates.
Then match what you see to a cause. This is our own triage table, built from the sources above. It is a starting point, not proof.
| What you see | Likely cause | Fix |
|---|---|---|
| Conversions far above form submissions | Duplicate tag, or a GA4 import plus a native tag, both primary | Keep one as primary, set the other to Secondary (observe only), add a transaction ID |
| Conversions cluster in office hours or on launch days | Test or staff submissions after clicking your own ads | Test with a separate form or a tag rule, retract known test conversions |
| Conversions with no matching CRM record and junk details | Bot form fills from paid clicks | Form checks and blocking repeat sources, as in the fake leads guide |
| Several conversions per lead | Every counting on a lead action, or thank-you page reloads | Switch the lead action to One |
| Conversions rising, qualified leads flat | Micro-conversion or soft event set as primary | Demote it to Secondary |
| Imported conversions you cannot tie to sessions | CRM import of every lead, view-through duplicates or a long view-through window | Import only qualified stages |
To put a cost on the gap, compare platform CPA with qualified CPA as in our guide to CPA inflation in Google Ads.
Which conversion actions should be primary, and Every or One?
Make only the action you want the bidder to chase primary, and count leads as One. Everything else can stay in the account as Secondary, where you still see it.
The setting is called Conversion action optimization, with two options: Primary and Secondary (observe only). Google says primary actions are used for bidding "as long as the standard goal they are part of is used for bidding" (About primary and secondary conversion actions, Google Ads Help). Secondary actions are "for observation only" and "not for bidding", with one exception Google notes: a custom goal that includes them. Demoting a doubtful action is the safest first move, because you keep its data and stop it steering bids.
Counting is the second setting. Under Every, "Google Ads counts every conversion (per tracked conversion action) that happens after an ad interaction." Under One, "Google Ads counts only one conversion per ad click" (About conversion counting options, Google Ads Help). Google pairs One with leads, where you care that a lead happened rather than how many sales followed, and Every with sales, where each purchase adds value.
How long should the conversion window be?
The conversion window is how long after an ad interaction a conversion can still count, and Google sets three. The click-through window covers conversions after a click, with a default of 30 days for Search and Display campaigns. The engaged-view window covers conversions after a video engagement, with a default of 3 days. The view-through window covers conversions after an ad was seen but not clicked, with a default of 1 day (About conversion windows, Google Ads Help). For lead actions, match the click-through window to how long your buyers really take to enquire, and keep view-through short. A long view-through window credits leads to people who only saw an ad.
A traced example
This is illustrative, not data from an account. Take one real enquiry. The form fires the native Google Ads lead tag, and GA4 imports the same form as a key event. Both are primary. The action counts Every, and the visitor reloads the thank-you page once. One person, one lead, and the Conversions column can show more than one conversion for it. Set the GA4 import to Secondary (observe only), switch the lead action to One, and the same enquiry counts once.
Who this is wrong for: an ecommerce account with repeat purchases should keep Every on its purchase action. A second order from the same click is a real second sale. Apply One to lead actions, not to everything.
Should you use data exclusions for fake conversions?
Data exclusions are for conversion data outages, not spam, and they do not change conversion reporting. Google says "Data exclusions are recommended for a conversion data outage, which includes conversion tracking issues or conversion uploads" (Use data exclusions for conversion data outages, Google Ads Help). Ongoing junk leads are not an outage.
A duplicate or broken tag that ran over a known period can qualify. The same page lists an "Incorrectly placed conversion tag on a website resulting in inaccurate conversion counts" as an example where data exclusions apply. Even then, "Data exclusions don't affect conversion reporting" (About data exclusions, Google Ads Help). An exclusion changes what bidding learns from. It never cleans the Conversions column. For the click-date rule and how many clicks to exclude, see the data-exclusions section of our fake leads guide, linked above.
How do you correct conversions that already counted?
Retract them with conversion adjustments where you have the order ID or GCLID, otherwise fix the action and let the bad period age out. A retraction is the only tool here that changes your reports. Google allows adjustments within 54 days, and within 7 days if you want bidding to use them.
Google's adjustments page says you can adjust a conversion "either to restate the value of the conversion or to retract it completely from your reports" (How to adjust your conversions, Google Ads Help). For website conversions it needs the order ID you sent with the tag. For imported offline conversions you can use the GCLID with the conversion time. Google states that "conversion adjustments can be made within 54 days". The shorter bidding window runs up to 7 days after the conversion is first recorded, so retract fast. Google also notes that website conversions cannot be added back to reports once removed, so check the list before you upload it.
That is the contrast with data exclusions: a retraction changes reporting, and an exclusion does not.
Google does some cleanup itself. Its invalid traffic page says "Google removes some invalid conversions if they originate from invalid traffic", and "This removal can occur after conversions appear in your account, which might show as a small drop in conversions over time" (About invalid traffic, Google Ads Help). If your totals dip slightly for past dates, that may be the reason. Google's page does not say it removes duplicates or test fills.
Run this conversion action cleanup checklist
Work through these ten steps in order, once now and again after any tag or form change. This is our working procedure, built from the Google pages cited above.
- 1. List every primary action. Goals, then Conversions, then Summary.
- 2. Match each to a real customer action. A lead, a call you can verify or a purchase.
- 3. Demote the rest. Set them to Secondary (observe only).
- 4. Find doubles. One event tracked by two tags, or by a tag plus a GA4 import.
- 5. Add transaction IDs. Wherever the form or checkout can pass one.
- 6. Set lead actions to One. Keep Every for purchases.
- 7. Never test after clicking your own ads. Use a tag rule that blocks test fills.
- 8. Audit imports. Import qualified stages, not every CRM record.
- 9. Retract known junk. Use adjustments within 54 days: the order ID for website conversions, or the GCLID with the conversion time for imports.
- 10. Use a data exclusion only for an outage. A broken or duplicated tag over known dates.
What does ClickFortify change, and what can't it see?
ClickFortify reduces one of the five sources, bot form fills from paid clicks, and nothing else on this list.
ClickFortify scores every Google Ads click on 100+ device, network and behavior data points in real time and excludes invalid IPs automatically through your Google Ads IP exclusion lists, with an evidence trail for each block. Fewer fraudulent clicks reach your forms, so less junk reaches your conversion data. See how it works for Google Search protection.
The limits matter as much. ClickFortify does not edit, filter or remove conversions in Google Ads, and it does not change your conversion actions, counting settings or imports. It cannot see lead form asset submissions, call-asset calls or Local Services Ads leads, because those never load your website. It works on the landing-page path only. For spam inside lead form assets, see our guide to Google Ads lead form spam.
Duplicate tags, test traffic, counting and imports are yours to fix with the checklist above. Our verdict: audit your conversion actions before you buy any protection, including ours. If the column still runs ahead of your CRM after the cleanup, the remaining gap is more likely to be traffic, and that is the part click protection addresses.
