Contents
- A useful funnel has four steps, not eight: beyond that, each extra step adds noise without adding a decision.
- GA4's funnel exploration takes about twenty minutes to build, provided you know which events exist before you start.
- A funnel does not tell you why people leave, it tells you where. The cause is found elsewhere.
The problem with default explorations
GA4's path exploration displays everything it finds: every event, every branch, every intermediate page. The result is a dense graph, technically correct and impossible to talk through in a meeting.
A conversion funnel exists to answer one question: between arrival and purchase, where do we lose the most people? Anything that does not serve that question is clutter.
The four steps that are enough
For an online store, the funnel that works everywhere is this one.
- Arrival: session started, or a category page view depending on the real entry point.
- Interest: product page view.
- Intent: add to cart.
- Purchase: order confirmed.
For a services site, the structure is identical with different events: arrival, offer page view, form opened, form submitted.
It is tempting to add checkout started, details entered, confirmation page. With each extra step the reading gets harder and the rates get too small to be meaningful. Keep the detail for a second exploration, once you know where to dig.
Building the exploration
In GA4, open Explore then choose the funnel exploration type. Add your four steps by selecting the matching events, in order.
Two settings change everything. Open or closed funnel: closed, only users passing through every step in order are counted; open, those who join partway through are counted too. Start closed, it reads more clearly. And breakdown: add the default channel group to see whether the drop-off comes from one specific source.
Reading the result
Look at the step-to-step completion rates first, not the absolute volumes. The biggest drop-off is not always where you expect: on many sites the main loss sits between arrival and the product page, not at checkout.
Then compare those rates by channel. An overall conversion rate of 2% can hide 5% on brand traffic and 0.4% on an acquisition campaign, which calls for two opposite decisions.
And keep in mind what the funnel does not say. It locates the loss, it explains nothing. The cause is found in session recordings, load times, or simply by walking the journey yourself on a phone.
The mistake to avoid
Comparing this funnel to Google Ads or Meta figures. The denominators and attribution windows differ, the gaps are structural, and the conversation goes off the rails. The subject is covered in our piece on what cannot be compared between Meta Ads and Google Ads.
A GA4 funnel is read against itself, from one period to the next. That is its only honest comparison.
Frequently asked questions
The Adwise team connects Google Ads, Meta Ads, LinkedIn Ads, GA4 and Search Console so agencies and freelancers can run their campaigns from a single place.