In 2026, a large share of Google Ads accounts that report a sudden jump in conversions are not winning — they are double-counting, and Smart Bidding is now optimizing toward a fantasy. When the dashboard shows more conversions than you have real sales, every automated decision downstream inherits the lie: bids climb on traffic that never pays back, and your true cost per sale rises while the reported numbers look better than ever.
This guide works through the five mechanical causes of inflated conversions — a duplicate tag, the wrong counting setting, a GA4 import overlap, thank-you page reloads, and the cleanup that protects your history — so you fix the overcount instead of trusting it. To check your account against the most common tracking faults automatically, run our free 5-axis Google Ads audit.
Updated 2026-05-16 with current conversion-counting, GA4 import and tag-deduplication behavior observed across US, UK and European accounts.
- Confirm against real sales — compare Google Ads to CRM orders; a roughly 2x gap is a duplicate. 2. Duplicate tag or trigger — the same event firing twice doubles almost everything. 3. Wrong counting setting — 'count every' on a lead form turns one person into many. 4. GA4 import overlap — a native tag and an imported key event both count one action. 5. Reload firing — a refreshable thank-you page logs the same order again and again.
How do you confirm conversions are inflated vs real?
Before you touch a tag, prove the count is wrong. Inflation is dangerous precisely because the dashboard looks healthy — more conversions read as success, so the problem hides in plain sight until your real cost per sale quietly climbs.
The reconciliation — Pull Google Ads conversions for the exact window of your CRM orders or back-office sales, and compare the two. If Google Ads shows materially more conversions than you have actual orders, the count is inflated. This single comparison is the fastest, most reliable confirmation you can run, and it costs nothing.
The ratio tells the cause — The shape of the gap points straight at the mechanism. A count running close to exactly 2x real sales almost always means a duplicate tag firing twice on every conversion. An uneven gap that spikes only on certain sessions usually means a wrong counting setting catching repeat submissions. Note the ratio before you start.
Rule out attribution noise — A modest gap can come from view-through conversions, cross-device, or modeled conversions, which are legitimate. The signal you are hunting is a large, structural gap — not a few percent. This is the mirror image of undercounting or zero conversions; if instead the two tools merely disagree, our GA4 and Google Ads discrepancy guide reconciles them properly.
Is a duplicate tag or GTM trigger firing twice?
The most common cause of an exact doubling is a conversion that fires twice on a single page load. Two tracking snippets, or one snippet plus an overlapping GTM trigger, both record the same action — and Google Ads dutifully counts both.
Two snippets on one page — A classic pattern is a hardcoded Google Ads conversion tag left in the template after the team also added the same conversion through GTM. Both fire on the thank-you page, so every order becomes two conversions. Search the page source and your tag manager for the same conversion ID appearing twice.
Overlapping GTM triggers — Inside Google Tag Manager, two triggers can match the same event — for example a page-view trigger and a custom-event trigger that both catch the confirmation step. The tag fires once per trigger, doubling the count. Audit your triggers so exactly one path fires each conversion.
Watch it live — Open the thank-you page with Google Tag Assistant or GTM preview mode and reload once. If you see the conversion event appear twice on a single load, you have found your duplicate. For a durable setup that avoids these races, our server-side GTM guide centralizes firing so an action can only be counted once.
Is 'count every' wrong for your conversion type?
Not every overcount is a broken tag. Sometimes the tag is firing correctly and the counting setting is simply wrong for the kind of action, so Google Ads inflates by design.
Count one vs count every — The 'count one' setting records a single conversion per ad click no matter how many times the action fires, while 'count every' records all of them. For a purchase, 'count every' is correct: 3 orders in a session are 3 real sales. For a lead form, signup, or quote request, 'count one' is correct, because the same person submitting twice is one intent, not two.
The classic inflation — A lead-gen account left on 'count every' is a textbook overcount. A visitor who hits submit, sees a slow response, and submits again logs two or three conversions for one lead. Across thousands of sessions this silently inflates the whole account and biases every bid.
Match setting to value — The test is simple: would a second event from the same person in the same visit be real, separate value, or the same intent counted twice? If separate value, use 'count every'; if the same intent, use 'count one'. A clean tracking foundation, covered in our enhanced conversions setup guide, makes these settings far easier to audit.
Is a GA4 import double-counting the same action?
A subtler cause appears when the same conversion is recorded by two systems at once. GA4 imports are the usual culprit: an action already tracked by a native Google Ads tag is also imported as a GA4 key event, so both count it.
The overlap — If you import a GA4 purchase key event into Google Ads while a native Google Ads purchase tag still fires on the same checkout, the order is counted twice — once by each source. This often happens after a migration, when the GA4 import is added but the old native tag is never removed.
One source of truth — The rule is one source per action: either the native tag or the GA4 import counts a given conversion, never both. Decide which one you trust, mark it as the conversion action, and stop the other from counting. Two GA4 key events describing one action cause the same problem.
Audit primary actions — Review which conversion actions are set as primary and feeding bidding. If a native tag and a GA4 import for the same event are both primary, you are double-counting into Smart Bidding. Removing the overlap — not deleting history — fixes the inflation at the source.
Are refresh, back-button or thank-you reloads firing it?
Even a single, correct tag can overcount if the page it sits on can be reloaded. The thank-you page is the danger zone, because customers reload, bookmark, and navigate back to it far more than you would expect.
Reload firing — When the conversion tag lives on a confirmation URL the visitor can refresh, every reload fires it again. Browser back-and-forward, a bookmarked return, or simply hitting refresh out of habit each adds a conversion. On a 'count every' purchase action, one happy customer can log several conversions for a single order.
Deduplication — The fix is to pass an order or transaction ID with each conversion and enable deduplication, so Google Ads ignores any repeat fire carrying an ID it has already seen. The same transaction can then only ever count once, no matter how many times the page loads.
Server-side reliability — Firing the conversion server-side, from the confirmed order rather than the browser, removes reloads from the equation entirely because the event is tied to the order record, not the page view. The honest count this produces is the foundation Smart Bidding needs to bid on real value instead of phantom volume.
The double-counting diagnostic table
Work this table top to bottom — it is ordered by how fast each cause is to confirm and how often it is the real reason Google Ads conversions outrun real sales.
Deleting a conversion action to stop the overcount wipes the history Smart Bidding learned on and can throw bidding into a hard reset. Instead, switch the action to secondary or fix its counting and source, so the inflated numbers stop accruing while the historical record stays intact for comparison. Expect the reported count to fall 20 to 40 percent toward real sales after the fix — that drop is the correction working, not a new problem. Let bidding re-learn on the honest signal before you judge any campaign.
How do you fix counts without losing history?
The fixes above usually resolve more than one cause at once, because inflated accounts often carry a duplicate tag and a wrong counting setting together. Work them in order — confirm the overcount, remove duplicates, correct counting, end any GA4 overlap, deduplicate reloads — and re-measure after each change rather than all at once, so you can see which lever moved the number.
The hardest part is psychological, not technical: a falling conversion count feels like a step backward when it is actually the moment your data starts telling the truth. Once the count matches real sales, Smart Bidding finally optimizes toward real value, and the cost per sale you see on the dashboard is the cost per sale you actually pay. To surface every counting fault automatically, run the SteerAds free 5-axis audit, and to confirm your tags fire exactly once on the right URL, use our GA4 URL tester.
Sources
Official sources consulted for this guide:
-
support.google.com — about conversion counting
-
developers.google.com — Google tag platform
-
support.google.com — about duplicate conversion tracking
-
ads.google.com — Google Ads
FAQ
Why does Google Ads show more conversions than sales?
In 2026 the overcount almost always traces to one of four mechanical causes, not to fraud or a magic boost. First, a duplicate tag: the same conversion fires twice because two snippets or two GTM triggers cover the same page. Second, the counting setting: 'count every' on a lead form turns one person into many conversions. Third, a GA4 import that overlaps an action Google Ads already records natively, so both count it. Fourth, a thank-you page that fires again on refresh or back-button. Compare Google Ads conversions to your CRM orders, and the gap names the cause.
Count every vs count one — which should I use?
Use 'count one' for actions where a person can only meaningfully convert once per visit, such as a lead form, a signup or a quote request, because counting every submission inflates a single intent into several. Use 'count every' for purchases, where each transaction is a distinct sale you genuinely want to count, so 3 orders in one session should read as 3 conversions. The classic inflation bug is a lead-gen account left on 'count every' while a frustrated visitor submits the same form three times. Match the setting to whether repeat events are real value or the same person twice.
Does a GA4 import double-count conversions?
It can, and this is one of the most common 2026 causes. If you import a GA4 key event into Google Ads while the same action is also tracked by a native Google Ads tag, both fire on the same purchase and the conversion is counted twice. The same happens when two GA4 key events describe one action. The rule is one source of truth per action: either the native Google Ads tag or the GA4 import counts a given conversion, never both. Check which conversion actions are marked primary and remove the overlap rather than deleting history.
Can a page refresh inflate conversions?
Yes, and the thank-you page is where it happens. If the conversion tag sits on a confirmation URL that the visitor can reload, every refresh, every browser back-and-forward, and every bookmarked return fires the tag again. A 'count every' purchase action then logs 2, 3 or more conversions for a single order. Server-side or order-ID deduplication stops it by ignoring repeat fires of the same transaction ID. Until that is in place, a single grateful customer hitting refresh on the thank-you page can quietly add a fake conversion every time.
Will fixing this drop my reported conversions?
Yes, and that is the point — your numbers will fall toward the truth. Once you remove a duplicate tag, switch a lead action to 'count one', or end a GA4 overlap, the inflated count drops to match real sales, often by 20 to 40 percent. That feels like a loss but it is a correction: you were optimizing Smart Bidding on phantom conversions. Expect a short re-learning window after the fix while bidding recalibrates to honest volume. Keep the old conversion action visible rather than deleting it, so historical reporting stays intact for comparison.
How do I know if it is a tag or the counting setting?
Look at the shape of the inflation. A duplicate tag tends to double almost everything — your conversions run close to exactly 2x real sales across the board. A wrong 'count every' setting inflates unevenly, spiking only on sessions where someone submitted or refreshed multiple times. Use Google Tag Assistant or the GTM preview to watch the page: if you see the conversion event fire twice on one load, it is a duplicate tag. If it fires once per submission but you get many submissions per person, it is the counting setting. The pattern points you straight at the cause.
Does Smart Bidding get worse with inflated conversions?
Yes, because automated bidding trusts the conversion signal completely and cannot tell a phantom from a real sale. If your count is inflated 2x, Target CPA and Target ROAS believe each click is worth twice its true value, so they bid up on traffic that does not pay back and your real cost per sale climbs while the dashboard looks great. The damage is largest on value-based strategies, which compound the error across the auction. Clean the conversion signal first, then let bidding re-learn on honest data before you judge any campaign.