GA4 vs Google Ads conversions: why the numbers don't match, and how to fix it

It's one of the most common questions in paid search: Google Ads says 42 leads this month, GA4 says 31, or the other way round, so one of them must be wrong. Usually neither is. The two tools answer different questions, on different dates, with different rules about who gets the credit.

The gap still matters. Google Ads bids with its own numbers, while you probably judge the business on GA4's. Below are the main documented reasons the two disagree, which ones you can fix, and a routine for checking that tracking isn't simply broken. Every statement about how the two tools work comes from Google's own help pages, checked on 12 October 2026; where something is our advice, we say so.

First, the words changed

On 27 March 2024 Google renamed GA4's "conversions" to key events. A "conversion" now means an action you use to measure ad campaigns and optimise bidding, and it means the same thing in GA4 and Google Ads. So the chain is: an event (any tracked action), a key event (an event you've marked as important in GA4), and a conversion (a key event or tag that Google Ads counts).

You can feed Google Ads in two ways, and you can use both. Conversions imported from GA4 can include results from channels other than Google Ads. The Google Ads conversion tag, installed directly or through Google Tag Manager, measures only conversions from Google Ads.

The settings that make the numbers disagree

Most of the gap comes from six settings that each tool keeps separately. This is the first table to check:

Google notes that for conversions created from GA4 events, the counting method and the conversion windows match by default. The attribution channels, the model and the time zone can still differ.

Click date versus conversion date

This one causes the most confusion. Google's own example: someone clicks your ad on 19 July and buys the next day, 20 July. Google Ads records the conversion on 19 July, the day of the click. GA4 records it on 20 July, the day of the purchase.

For a business where people enquire weeks after the first click, this shifts conversions between weeks and months. It also makes the most recent days look bad in Google Ads, because conversions from recent clicks haven't happened yet. Cost per lead looks high this week and quietly improves next week.

The fix takes thirty seconds. In Google Ads, open Columns and add Conversions (by conv. time), or All conv. (by conv. time), which also includes secondary actions. Those columns put each conversion on the day it happened, like GA4. In GA4's Conversion performance report you can switch between conversion time and interaction time too.

Which traffic gets the credit

GA4 counts a key event whichever way the visitor arrived: organic search, Instagram, a WhatsApp link, typing your address. Google Ads only reports conversions that follow an interaction with an ad in the same Google Ads account. So if you compare GA4's total key events with Google Ads, GA4 will usually be higher, and that tells you nothing about tracking quality.

GA4 also decides which channels can receive credit when it shares data with Google Ads. You choose between Google paid channels and Paid and organic channels. New links have defaulted to Google paid channels since June 2023. With paid and organic, GA4 can give part of the credit to organic search or email, and only the share it gives to Google Ads appears in Google Ads.

To compare like with like, use GA4's Advertising section, or filter a report to the source/medium google / cpc. Don't compare Google Ads with GA4's Traffic acquisition totals.

Counting, windows and attribution

Counting method

If one person fills in your enquiry form twice, in different sessions, GA4 counts two key events. Google Ads counts one if the action is set to "One" and may count two if it's set to "Every". "Every" is the default for website actions. For lead generation, set Google Ads to One, because a second form from the same person isn't a second customer. For sales, use Every and send a transaction ID with each purchase so the same sale isn't counted twice.

Conversion windows

Google Ads counts conversions within a window after the interaction: 1 to 90 days, 30 days after a click by default, 3 days for engaged video views and 1 day for view-through. Google Ads cookies expire 90 days after the click, while the GA4 cookie can last up to two years. Someone who clicks an ad and enquires four months later can still be counted as a key event in GA4 (credited to another channel), but will never show in Google Ads. Since 11 August 2026, conversions created from GA4 also accept custom windows.

Attribution model

Google Ads now offers two models: data-driven, the default for most actions, and last click. The older rule-based models (first click, linear, time decay, position-based) were retired, and actions using them were moved to data-driven. Data-driven attribution splits credit across several clicks, which is why you sometimes see 0.4 of a conversion. GA4 sets its reporting model for the whole property, and a change applies to historical data too.

When you can trust the numbers

Both tools need time. Conversions imported from GA4 reach Google Ads up to 24 hours after they happen. GA4's own processing can take 24 to 48 hours. And GA4 can keep changing how it credits a key event to channels for up to 12 days while its models update.

The practical rule: compare periods that ended at least 12 days ago, and never judge yesterday. If you bid on imported GA4 key events, Smart Bidding is also working with data that is up to a day late.

Consent banners, cookies and lost click IDs

These are where real data goes missing, rather than being counted differently.

Consent mode: basic or advanced

In basic consent mode, Google's tags don't load until the visitor answers the cookie banner, and nothing at all is sent if they decline. In advanced mode the tags load and send cookieless pings while consent is denied, which lets Google model the conversions it can't observe with a model specific to your account. Basic mode gets only a general model.

Modelling has thresholds. Google Ads needs 700 ad clicks over 7 days, per country and domain grouping, before it models consented-out conversions. GA4's behavioural modelling needs advanced mode on every page, at least 1,000 events a day from visitors who declined analytics for 7 days, and at least 1,000 daily users who accepted it on 7 of the previous 28 days. A business with fewer than 700 ad clicks a week won't qualify for Google Ads modelling, and GA4's behavioural modelling estimates users and sessions, not key-event counts. So for most small accounts a declined banner usually means a lead neither tool counts.

Two common mistakes: the banner blocks the tags entirely instead of using consent mode, or the site sets consent with a "default" command after the visitor chooses instead of an "update". Google lists the second as a cause of "(not set)" traffic in GA4. Our guide to DPDP consent for Indian websites covers what the banner itself should say.

The GCLID has to survive

Auto-tagging adds a click ID called the GCLID to every ad click. It's on by default for new accounts and it's required to import GA4 key events. Redirects (301, 302 or JavaScript), URL rewrites that change letter case, truncated URLs and iframes can all lose it. Then Google Ads can't match the conversion, and GA4 may file the visit under organic, direct or "(not set)".

Google's test: paste your landing page URL into Chrome, add ?gclid=TeSter-123 to the end (use & instead of ? if the URL already has a ?, and put it before any #), load it, and check that the parameter is still in the address bar after any redirects. Google also gives a much longer test value to check that your site doesn't truncate long click IDs.

Conversions that only one tool can see

• View-through conversions. Someone sees a Display or YouTube ad, doesn't click, and converts later. For most campaign types these sit in Google Ads' "View-through conv." and "All conv." columns, not in "Conversions". Conversions imported from GA4 don't get view-through credit at all, because that needs Google Ads' own measurement.
• Engaged-view conversions. Someone watches at least 10 seconds of a skippable YouTube ad (or the whole ad if shorter), or 5 seconds of an in-feed or Shorts ad, then converts within the window. These are included in Google Ads' "Conversions" column.
• Cross-device and modelled conversions. Google Ads models conversions where someone clicks on a phone and enquires on a laptop. For conversions imported from GA4, cross-device credit needs Google signals switched on in GA4.
• Invalid clicks. Conversions that GA4 records from clicks Google later judges invalid can be filtered out on import, so Google Ads can be slightly lower and can drop a little after the fact.

Import GA4 key events or use the Google Ads tag?

Google lets you use either, or both. Three documented facts decide it for most businesses:

• Conversions created from GA4 key events are set to secondary by default when the account already has goals from a GA4 property, "to prevent counting the same event twice for bidding". Only primary actions feed the Conversions column and bidding (unless a secondary action is part of a custom goal). You can only make them primary inside Google Ads.
• Google's enhanced conversions setup page says conversions imported from Google Analytics goals aren't supported, and suggests a Google Ads conversion action with the Google tag or Tag Manager instead.
• Imported GA4 conversions get no view-through credit, and they arrive later (12 to 24 hours against 3 to 15 for the tag).

Our view, for businesses that want enquiries: make the Google Ads tag, set up through Google Tag Manager with enhanced conversions, the primary action Smart Bidding learns from. Keep the GA4 import as a secondary action so you can still compare. Never make two actions that count the same form both primary, or Google Ads will count each enquiry twice and bid as if leads were twice as cheap as they are.

Tag fires but Google Ads shows no conversions: a checklist

If Google Ads shows zero while GA4 shows leads, something is broken rather than counted differently. Work down this list in order:

1. Open Goals, then Conversions, then Summary in Google Ads and read the status. "Awaiting conversions" is normal for an action less than 48 hours old or a paused campaign; "Unverified" is normal until the first conversion or a successful test.
2. Click Troubleshoot to open Tag Assistant, and start from the landing page, not the thank-you page. With Tag Manager, choose to debug the live version. Google says verified actions update in Google Ads after about 30 minutes.
3. Check that auto-tagging is on, then run the ?gclid=TeSter-123 test above.
4. In Tag Manager, check that a Conversion Linker tag fires on all pages. A container that loads a Google tag on every page doesn't need a separate one.
5. Check that the conversion ID and label in your tag match the action in Google Ads, and remove old or duplicate tags and extra containers.
6. Check the cookie banner: tags should load under consent mode, and the update command must run after the visitor's choice.
7. If the conversion happens on another domain (a booking tool, a payment page), set up cross-domain measurement. Don't fire conversion tags inside iframes.
8. For imported GA4 key events: confirm the event is marked as a key event, that the GA4 property is linked to the account that owns the conversion action, and that there is recent google / cpc traffic.
9. Check the action is primary and included in the campaign's goals, or add the "All conv." column to see secondary actions.
10. After a fix, wait. Google documents 24 to 48 hours for tag status to update, and reporting can lag 15 to 48 hours for conversions.

A monthly reconciliation that takes ten minutes

A routine you can run every month: pick a full month that ended at least 12 days ago. In Google Ads, use "Conversions (by conv. time)" for your lead action. In GA4, open Advertising and look at the same key event for Google Ads traffic over the same dates, using the same attribution model. Then compare both with the leads you actually received in your inbox, CRM or WhatsApp.

If the two tools are within about 20% of each other and both are close to your real lead count, tracking is healthy. If Google Ads is far higher than your real leads, look for duplicate primary actions. If both are far lower, look at the consent banner and the GCLID first.

How we set up conversion tracking

Our one-time tracking setup is ₹10,000. It covers a written measurement plan, a Google Tag Manager container and data layer, GA4 key events and Google Ads conversions, the Meta Pixel with the Conversions API and deduplication, Consent Mode v2 with a banner, and a Looker Studio dashboard. We test every form, WhatsApp and call click before handing it over.

See what's included on the conversion tracking and analytics page. If your numbers already disagree, send us screenshots of both. We'll tell you which of the reasons above explains the gap.