- In 2024, GA4 renamed conversions to key events. A key event marks an action that matters to the business in GA4 reports. A conversion now means a Google Ads conversion used to measure ads and guide bidding.
- Most stores should mark very few key events: purchase, plus one or two real business outcomes such as an email signup. Funnel steps like view_item and add_to_cart stay plain events.
- GA4 can mark extra events on its own, including add_to_cart and session_start, when Ads Personalization is on and Google Ads is linked. Audit the list, then feed Google Ads bidding from purchase with value only.
What is a GA4 key event, and what changed in 2024?
A key event is an event you mark in Google Analytics 4 because it measures an action that matters to your business. Any event GA4 collects can be marked. Once marked, it appears in the key events metric, in key event rates and in attribution reports.
Before 2024, these were called conversions in GA4. The trouble was that “conversions” in GA4 and “conversions” in Google Ads were measured differently, so the same word showed different numbers in two tools. Google split the terms:
- Key event: an important action, reported in GA4.
- Conversion: an action used to measure ad performance and optimize bidding in Google Ads. You can create a Google Ads conversion from a GA4 key event, and the two platforms share it.
Your history did not disappear. Data that was collected as conversions now shows as key events, so year-over-year comparisons still work.
| GA4 key event | Google Ads conversion | |
|---|---|---|
| Where you set it | GA4 Admin, Events or Key events | Google Ads, or GA4 for linked accounts |
| What it is for | Reporting what matters in GA4 | Measuring ads and guiding bidding |
| Where results show | GA4 reports and explorations | Google Ads, and the Advertising section of GA4 |
| Typical eCommerce use | purchase, email signup | purchase with value |
| Limit | 30 per standard property, 50 for Analytics 360 | Set by Google Ads configuration limits |
One detail that confuses teams: Google Ads conversions do not appear in standard GA4 reports. They show in Google Ads and in GA4’s Advertising section. That is one reason the numbers in the two tools rarely match exactly.
Which events should an eCommerce store mark as key events?
Mark the outcomes you would report to a founder. Leave the steps on the way there as plain events.
The test is simple: if this number went up next month and nothing else changed, would the business be better off? More purchases, yes. More email subscribers, usually yes. More product views or more sessions, not necessarily.
| Event | Mark as key event? | Why |
|---|---|---|
| purchase | Yes | The outcome. GA4 marks it by default on web |
| Email or SMS signup (for example, sign_up or a custom email_signup) | Yes, if it feeds your lifecycle program | A real asset you can market to |
| generate_lead (wholesale or B2B form) | Yes, if you sell wholesale | A sales outcome with its own value |
| Subscription start (custom event) | Yes, for subscription stores | Different value from a one-off order |
| begin_checkout | Usually no | A funnel step; analyze it in funnel reports |
| add_to_cart | No | High volume, weak link to revenue |
| view_item, view_item_list | No | Browsing, not an outcome |
| view_search_results | No | Useful for search analysis, not a business result |
| session_start, page_view, scroll, click | No | Inflates key events without meaning |
Funnel events are still valuable. You need view_item, add_to_cart and begin_checkout to find where shoppers drop off. They just do not need to be key events to do that job. You can build a full funnel from plain events, as our guide to GA4 funnel exploration shows.
Why marking too many events hurts
Every marked event is added to the key events metric and the key event rate. Mark add_to_cart and session_start, and a channel that sends lots of browsers looks as good as a channel that sends buyers. Your attribution reports start rewarding cheap actions. The key event rate turns into a number nobody can explain.
Why are there key events you never marked?
GA4 marks some events as key events by default. On web, that is purchase. Google’s help center also lists app events such as first_open and in_app_purchase.
There is a second list that surprises many stores. According to Google’s help center, when you enable Ads Personalization with a linked Google Ads account, GA4 also marks these as key events:
- add_to_cart
- add_to_wishlist
- begin_checkout
- session_start
- view_item
- view_item_list
- view_search_results
If your Google Ads account is linked and personalization is on, check your Key events list. You may find most of your funnel there, inflating every key event figure. Before you unmark any of them, check whether a Google Ads conversion or an audience depends on it, so you do not break bidding or remarketing.
How should key events connect to Google Ads?
Bidding should optimize toward the outcome that makes money, not toward activity.
- Use purchase with value as the primary conversion for Shopping, Search and Performance Max campaigns. The value parameter should carry the order value, so value-based bidding can work.
- Keep micro events secondary at most. An add-to-cart conversion set as primary teaches bidding to find people who add to cart, which is not the same as people who buy. Secondary conversions are observed but not used for bidding.
- Avoid double counting. If the Google Ads tag already records purchases, do not also import the GA4 purchase as a primary conversion. Pick one source as primary and set the other as secondary.
- Keep consent in the loop. In regions where you ask for consent, consent mode settings affect what both GA4 and Google Ads can record. Our guide to Consent Mode v2 covers the setup.
How do you mark, check and clean up key events?
Mark an event
In GA4 Admin, under Data display, open Events. Find the event in the Recent events tab and click the star next to it. You need Editor access or above to mark an existing event. To mark an event GA4 has not received yet, use Create event or add it on the Key events tab ahead of time.
New key events apply from the moment you mark them. They do not rewrite history. Google says to allow up to 24 hours for them to show in standard reports, while Realtime updates within minutes.
Choose the counting method
GA4 offers two counting methods for each key event:
- Once per event (recommended): counts every time the event fires. Use this for purchase, so two orders in one session count as two.
- Once per session (legacy): counts one per session. It is the default for key events that came from old Universal Analytics goals. It can suit a lead form that some people submit twice.
Check the purchase event itself
A key event is only as good as the event behind it. For purchase, confirm:
- A unique transaction_id on every order. Google uses it to help avoid duplicate purchase events, for example when a shopper reloads the thank-you page.
- value and currency on every order, with a consistent rule for tax and shipping.
- An items array with item IDs that match your product feed.
- One source of the event only. A theme snippet plus an app plus Google Tag Manager can send three purchases for one order.
If GA4 revenue does not match your store, our guide to the GA4 and Shopify revenue discrepancy walks through the usual causes.
Worked example (illustrative numbers)
All numbers below are illustrative.
A store’s Key events list contains eight events: purchase, add_to_cart, begin_checkout, view_item, view_item_list, view_search_results, session_start and email_signup. Most were added automatically after the Google Ads link.
In a month, the Traffic acquisition report shows:
- Paid social: 9,000 key events.
- Email: 2,000 key events.
Paid social looks four or five times stronger. But most of its key events are session_start and view_item from browsing visitors.
The team unmarks everything except purchase and email_signup, after checking that no Google Ads conversion or audience depends on the others. The following month, with similar traffic:
- Paid social: 320 key events (250 purchases, 70 signups).
- Email: 410 key events (400 purchases, 10 signups).
Now the reports show what the business cares about. Email drives more orders, and paid social is a strong source of new subscribers. Both facts were hidden under browsing activity. The cleanup did not change a single sale. It changed which channels looked like they earned them.
GA4 key events cleanup checklist
- [ ] Export the current Key events list and note who added each one.
- [ ] Check whether Ads Personalization and a Google Ads link have added funnel events.
- [ ] Keep purchase, plus one or two real outcomes such as an email signup.
- [ ] Before unmarking, check Google Ads conversions and audiences that use the event.
- [ ] Set purchase to count once per event.
- [ ] Confirm transaction_id, value, currency and items on every purchase.
- [ ] Confirm purchase is sent from one source only.
- [ ] Use purchase with value as the primary Google Ads conversion; keep micro events secondary.
- [ ] Avoid a second primary purchase conversion from another tag.
- [ ] Document each key event, its definition and its owner.
- [ ] Review the list every quarter and after any app, theme or tracking change.
FAQ
Does unmarking a key event delete my historical data?
No. Google states that turning off a key event does not affect data already collected, and past events still show in reports as key events. The change only applies from that point forward, so note the date in your reporting.
Should add_to_cart be a Google Ads conversion?
Not as a primary conversion for a store that has enough purchases for bidding. It teaches the system to find carts, not buyers. A store with very few purchases sometimes uses it temporarily, but set it as secondary as soon as purchase volume allows.
Why don’t GA4 key events match Google Ads conversions?
The tools use different attribution, counting rules and timing. Google Ads credits conversions to the click date and uses its own attribution model, while GA4 reports on the event date across all channels. Expect differences, and check that both sides use the same purchase event and counting method.