Short answer Google Ads and GA4 date conversions differently, credit them differently, and count them under different settings, so their totals differ even when the tracking works. Line up the conversion, the dates, and the columns first. If a gap is still there, or it jumped on a specific day, look for lost
gclids, tags that stopped firing, and renamed or duplicated events.
A difference between Google Ads and GA4 is not a bug report. A difference you can’t explain, or one that changed overnight, is.
First, know which setup you are comparing
There are two common setups, and they fail in different ways.
- Google Ads tag. Google Ads counts conversions from its own tag; GA4 counts its key events from its own. Two tags on the page, so a gap can come from the reporting rules or from one tag firing where the other doesn’t.
- GA4 key events imported into Google Ads. Both reports start from the same GA4 event. A gap here is almost always about dates, credit, timing, and settings, not about the tag.
Check which one each conversion action uses before you start. It decides where you look.
Why the totals differ when tracking works
Different dates. Google Ads reports a conversion on the date and time of the click that led to it; Analytics on the day the conversion happened. A purchase on the 1st after a click on the 28th lands in different months. Add the “Conversions (by conv. time)” columns in Google Ads to compare closer to GA4.
Different windows. A Google Ads conversion window can be between 1 and 90 days, and its click cookies expire after 90 days, while the Analytics cookie lasts up to two years. GA4’s own lookback window is 90 days by default for most key events, 30 for acquisition key events.
Different credit. Google Ads only reports conversions with Google Ads interactions from the same account. GA4 assigns credit across all channels under its reporting attribution model, so part of the same conversions goes to organic search, email, or Meta. GA4’s setting for which channels can receive credit, Google paid channels or paid and organic channels, also affects the key events shared with Google Ads; Google warns that changing it can affect bidding and reporting there.
Different columns and counting.
- Each conversion action counts “One” or “Every” conversion per click.
- “All conversions” adds secondary actions and view-through conversions to the Conversions column.
- Secondary actions are for observation only and may not appear in default reports, and view-through conversions are not in the Conversions column by default.
Estimates. Google Ads models cross-device conversions and may include modeled conversions where Google can’t observe all of them. GA4’s behavioral modeling for Consent Mode is something else: it estimates users and sessions, not event counts, and shows only with the Blended reporting identity.
Invalid clicks. Google may remove conversions from clicks it later finds invalid, and some events reported in Analytics are filtered out when imported. GA4 keeps them.
Timing. Analytics key events are imported into Google Ads up to 24 hours after they occur, conversion data can take up to 7 days to fully appear after significant changes, and conversions can be reported up to 90 days after the click. Compare closed periods.
Line up the two reports
| In Google Ads | In GA4 |
|---|---|
| The conversion action, and whether it is primary | The key event it comes from, or the one that matches it |
| “Conversions (by conv. time)” | The same date range |
| The conversion window of that action | The lookback window and attribution model |
| Counting: One or Every | The key event’s counting method |
| Campaigns from one account | Rows where the source is Google Ads, under the same model |
Run both with the same dates and compare by day and by conversion action, not as one total. A gap that is steady across days is usually the rules. A gap that starts on one day is usually a change. To see what that gap looks like in a real account, Swishiy walks through one.
When the gap is a tracking problem
These are the causes worth fixing, because they change what is collected, not how it is reported.
- The
gcliddoesn’t reach the landing page. Auto-tagging addsgclidto the URL, and if your site redirects, thegclidmust be passed to the final landing page for conversions to be tracked. A redirect, a link shortener, or a page that strips parameters breaks the link between the click and the conversion. GA4 may then credit the visit to another source. - The tag stopped firing. A release removes the tag from the thank-you page, or a trigger changes. Google points to the conversion tracking status troubleshooter and Tag Assistant to find inactive tags, including tags installed through platforms such as Shopify.
- A renamed key event. A release renames the event, GA4 counts the new name, and the import stops. A renamed key event can take up to 24 hours to be reflected in Google Ads.
- Payment and checkout redirects. A visitor who returns from a payment provider can show up in GA4 as a referral from that provider, unless you list it as an unwanted referral. The Google Ads conversion keeps its click; GA4 loses the campaign.
- Values that disagree. The Google Ads tag sends the value with tax and GA4 without, or a non-purchase action carries a value. Google notes that values on non-purchase primary actions can inflate conversion value.
- Consent applied differently. If one tag waits for consent and the other doesn’t, they collect different shares of the same traffic. The fix is to apply the visitor’s choice to both, never to collect what they declined.
A five-step check
- Identify the setup of each conversion action: Google Ads tag or imported GA4 key event.
- Align the reports with the table above and compare by day and action.
- Reconcile a sample against the backend. Take ten orders from yesterday and check which ones each platform counted. The backend confirms the order happened; it does not decide which channel earned it.
- Inspect the hits for those orders: did the conversion tag fire, with which value, with a
gclidon the landing, with which consent state? - Match the date of the jump with what changed: a release, a GTM container version, a CMP update, a checkout change.
If step 2 closes the gap to a stable size, there is nothing to fix in the tags. Document the definitions and move on.
Where Trackingplan fits
Trackingplan doesn’t change how Google Ads or GA4 assign credit. It shows you whether the hits behind both numbers are right. It watches the conversion hits your site sends to Google Ads, GA4, and the other platforms, on real traffic, with their value, gclid, and consent state. Ask the agent why the platforms disagree and it compares both counts for the same period, finds the cause, and ends with the fix to ask for. With your ad accounts connected, it also reads Google Ads spend and reported conversions next to what your site sent.
For the same question across Meta and TikTok too, see why GA4 and ad platforms report different conversions.
The gap between Google Ads and GA4 won’t close. It only needs to be the one the settings explain.
Frequently asked questions
What is an acceptable discrepancy between Google Ads and GA4?
There is no universal threshold. It depends on your conversion windows, counting settings, consent rates, and how many conversions come from other channels. What matters is that the gap is explained and stable. A gap that jumps on a specific day is the warning sign.
Can I make Google Ads and GA4 show the same number?
No, but you can make them comparable: same conversion, same dates, the conversion-time columns in Google Ads, and the same window where the settings allow it. The remaining gap is how each one assigns credit.
Which number should I trust?
The one that fits the decision. Google Ads for bidding and comparing campaigns inside Google Ads; GA4 for comparing channels; the backend or CRM for how many orders or leads actually happened. Write that policy down once instead of debating it every month.
Should I compare Google Ads clicks with GA4 sessions?
Only as a direction. A click is an ad interaction and a session is GA4’s visit logic. They move together, but they are not the same unit, and a one-to-one comparison creates false alarms.