How do you analyse a conversion funnel with GA4?

A conversion funnel is the sequence of steps a visitor goes through before the final action: viewing a product, adding it to the basket, starting checkout, paying; or, on a services site, reading an offer page, opening the form, sending it. At each step, some visitors drop out. Measuring those drop-offs is the most direct way to increase your conversions without buying more traffic. In GA4, that is what the funnel exploration is for.

Prerequisite: an event for each step

GA4 builds a funnel from events or page views. For an online store, the e-commerce events (view_item, add_to_cart, begin_checkout, purchase) are enough. For a services site, plan for a form_start event (supplied by enhanced measurement) and generate_lead on submission. If a step has neither an event nor a distinct URL, it can't be measured: that is the first thing to check.

Building the exploration in 5 steps

1. Open Explore > Funnel exploration (or start from the "Funnel" template).
2. Define the steps : in the Settings panel, click the pencil next to "Steps". Each step is a condition: an event name (add_to_cart) or a page ("Page path contains /basket"). Four to six steps are enough.
3. Choose the type : a closed funnel (visitors have to go through the steps in order) for a purchase journey; an open funnel if people can enter at any step (a visitor landing directly on the basket from an email, for example).
4. Add a breakdown : the "Device category" or "Session default channel group" dimension shows the funnel by segment.
5. Set the period : at least 28 days to smooth out variations, within the limit of your data retention.

Reading the result

For each step, the exploration shows the number of users, the rate of progression to the next step and the drop-off rate. Here is a typical example for a store:

The step to work on is the one whose drop-off is clearly above the benchmarks: here, moving from starting checkout to entering payment details. The classic causes at that step are shipping costs discovered too late, being forced to create an account, or a form that is too long on mobile.

Segmenting to find the cause

An overall drop-off rate often hides a localised problem. Three breakdowns to test as a matter of course:

• Device : a 62% drop-off on mobile against 35% on desktop at the same step points to a mobile usability problem (the card field, the keyboard, a hidden button).
• Channel : if Google Ads visitors drop out twice as often as SEO visitors at the very first step, the landing page doesn't match the ad's promise.
• New vs returning : a mass drop-off of new visitors at payment suggests a lack of reassurance (reviews, guarantees, payment methods displayed).

To go further, the exploration's "Next steps" feature shows where the visitors who drop out go: back to the product page (doubt), to the delivery page (costs), or out of the site.

A lead funnel: the case of service sites

The same tool applies to a contact journey: service page > contact page (page_view) > form started (form_start) > submitted (generate_lead). A high drop-off between "started" and "submitted" signals a form that is too long, or fields rejecting valid values (phone number, postcode). Cutting a form from nine fields to five commonly lifts the submission rate by 20 to 40%.

The limits to know about

• Retention : explorations use detailed data, kept for 2 or 14 months depending on your setting (see setting GA4 up properly).
• Sampling : beyond 10 million events over the period, GA4 samples; an icon at the top of the report says so.
• Thresholds : with Google Signals, segments that are too small are hidden.
• Users, not sessions : the funnel counts distinct users; a visitor who needs three sessions to buy counts once.

Once you've identified the weak step, test one change at a time and measure again over 28 days. The conversion GA4 KPIs will tell you whether the change paid off, and the overall conversion rate will follow.