Conversion rate on a service business's website: what to count, and how Google Analytics works it out

Google Analytics 4 calls the actions you count key events. Google defines a key event as "an event that measures an action that's particularly important to the success of your business", and any event the site sends can be marked as one. A key event can then be made into a conversion for Google Ads, in Google's own order: event, key event, conversion.

A standard property can mark up to 30 events as key events. Marking one changes the reports from that moment on, and Google says it does not change historic data, so the key event figures start on the day it is marked.

GA4's list of metrics defines two key event rates, in these words:

The two rates answer different questions. The user rate is the share of the people who visited that did the thing you count; the session rate is the share of visits that included it. Google's definition of a session ends it, by default, after 30 minutes of inactivity, so one person can make several sessions.

Each key event also has a counting method. Once per event, which Google recommends, counts a key event every time it is triggered; once per session counts it once within a session. Google's example is a key event triggered five times in one session: five key events under the first method, one under the second. A change of method applies only to key events counted after it.

For a business that sells a service, the actions worth counting are the ones that start a job:

GA4's enhanced measurement can record forms by itself, with two events: form_start, "the first time a user interacts with a form in a session", and form_submit, "when the user submits a form". That is the press of the button. Google's description says nothing of whether the message arrived.

Google's own tutorial for setting up a key event counts the page a visitor sees after a form is accepted. It creates a generate_lead event from a page_view of a confirmation address, such as /contact-us-submitted, and marks that event as the key event. Set the form to show that page only once the enquiry has been accepted, and the count follows the enquiries.

Tag Manager's form submission trigger has a setting for the same purpose. Google says its Check Validation option sets the trigger "to fire only if the form is successfully sent", and that without it "the trigger will fire whenever a user attempts to submit the form".

Then test it with a real enquiry: send one through the form, check that it arrives, and check that one key event appears in GA4's Realtime report. Testing every form for real is one of the things a website build should include.

Analytics cookies need consent in Ireland: the Data Protection Commission's guidance answers "Do analytics cookies require consent?" with "Yes." What you can measure without consent covers the rule, and what stays available without it.

Google's consent mode passes each visitor's answer to Google's tags, in one of two forms. In basic consent mode the tags do not load until the visitor answers the banner, and when the visitor does not consent, Google says no data reaches it at all. In advanced consent mode the tags load straight away, and while consent is denied they send measurements without cookies, which Google uses to model the gaps.

The modelling has a floor. Google's behavioural modelling, which estimates the behaviour of visitors who declined, needs advanced consent mode, at least 1,000 events a day with analytics storage denied for at least 7 days, and at least 1,000 daily users with it granted on at least 7 of the previous 28 days, and Google says meeting those does not guarantee it. Without enough consented traffic, Google says, events from visitors who decline "aren't reported". Its modelling of key events is included only when Google is confident of the model; without enough traffic, modelled key events are not reported, or in Google Analytics are attributed to the Direct channel.

Where no modelling applies, GA4's rate is a rate among the visitors who agreed to analytics, and its count of enquiries can fall short of the enquiries you received. The count that needs no consent is your own: the enquiries in your inbox or your booking system for the month. Compare it with GA4's key events for the same month, and the difference is what the figures miss.

A rate is one count divided by another, so it moves when either moves. A fall can come from fewer enquiries, from more visits by people who were never going to enquire, from a form that stopped sending, from a tag that stopped firing, or from a change to the banner. Look at the count of enquiries in your own records before the rate.

This article gives no average conversion rate to compare against. A rate depends on what a site counts as a conversion and which visits it counts, so your own rate over time is the comparison that holds.

Your rate is also the figure that turns the price of a click into the price of an enquiry. The free break-even ad calculator takes what a job is worth, your margin and the share of enquiries you win, and works out the most an enquiry or a click can cost. If your enquiries come from phones, the website speed test guide explains how Google measures a page loading on one.