What is a key event (GA4)

In Google Analytics 4 everything that happens on a website or in an app is recorded as an event: a page view, a click on an outbound link, a form submission, a purchase. A key event is neither a separate kind of event nor a tag that starts a new measurement. It is a marker you place on an event that is already being collected, telling Analytics that this action carries more weight than the rest.

The rename away from the word conversion has practical consequences. In the Analytics reports you will see the key events metric and its rate. The word conversion now belongs to what lives in Google Ads: a conversion item created from an event or from a key event, used for bidding, for campaign optimisation and for building audiences. Google Ads conversions do not appear in the standard Analytics reports, and a key event on its own feeds no bidding strategy.

Marking an event as key changes neither the data collection nor a faulty implementation. If the event never reaches the server, the marker will not make it appear.

The path starts with collection. The event has to exist before you can mark it, whether it comes from enhanced measurement, from your tag container or from a rule created inside the property itself. Once it shows up in the event list in the admin area, a single toggle turns it into a key event.

From that moment Analytics counts it according to the configured counting method. There are two options. The first, once per event, counts every trigger and is the recommended one, because it reflects how people actually behave on the site. The second, once per session, counts at most one per session and is the legacy option, applied by default to key events migrated from Universal Analytics goals. Changing the method affects future data only: history is not recalculated.

The number of key events per property is capped. Standard properties allow thirty and 360 properties allow fifty, so the list is managed as a scarce resource rather than as a drawer everything falls into.

Getting the action through to Google Ads takes one more step. In conversion management the conversion item is created from the key event, a category is assigned to it, and from then on the data is shared between both platforms. The counting method and the conversion window are inherited from the Analytics property.

The distinction between the two words explains an argument that recurs in almost every account: the numbers in Analytics and the numbers in Google Ads do not match, and they are not supposed to.

The reasons are documented. Analytics attributes with a cross-channel last click model and assigns the action to the day it happened. Google Ads reports the conversion against the date and time of the click that caused it, not against the date of the action, so two reports covering the same date range are describing different things. On top of that come conversion modelling in Ads, which can raise the total above what Analytics observed, cross-device conversion estimates, and the absence of deduplication when measurement happens at account level across several Ads accounts.

Understanding this changes the conversation with management and with the client. Instead of hunting for a culprit in the implementation every time a fifteen percent gap shows up, you document which platform answers which question. Analytics describes behaviour on the site and the weight of each channel. Google Ads measures the return on ad spend and needs its own bookkeeping in order to bid.

• Mark only the actions someone will actually use to make a decision, not everything that can technically be measured.
• Check that the event fires correctly before marking it, using DebugView and the realtime report.
• Choose the counting method deliberately and note the date of the change, because history is not recalculated.
• Document in a table which key event feeds which Google Ads conversion, and with which category.
• Review the list every quarter and free up slots by unmarking key events nobody looks at any more.
• Explain to management and to the client, before they ask, why the Analytics and Ads figures are not going to match.

• Marking a badly implemented event as key. The marker does not fix the measurement, it only turns a doubtful number into a prominent metric.
• Marking page views or high volume automatic events, which inflate the rate and drain any meaning out of campaign comparisons.
• Assuming that marking the event already sends it to Google Ads, when the conversion still has to be created in conversion management.
• Switching the counting method halfway through an analysis period and then comparing against history as if nothing had happened.
• Using up the property limit with test key events that were never removed.