How a funnel exploration works
You define the steps of a journey in order — viewed the service page, started the form, submitted the enquiry — and GA4 counts how many people reached each one and how many did not carry on. The output is a set of bars that shrink from left to right, with the loss between each pair of steps shown as its own figure.
Two choices change the meaning of everything you see. An open funnel lets people enter at any step, so somebody who arrived straight on the checkout page still counts; a closed funnel only counts people who started at the beginning. Separately, the elapsed-time option shows how long people take between steps, which often explains a drop that looks like abandonment but is really a delay.
Once the shape is built you can break each step down by a dimension — device, channel, country, landing page — and you can build a segment out of any group that fell out at a particular step. That last feature is the one people underuse, because it turns an observation into a list of pages to fix or an audience to re-approach.
Why funnel exploration matters
A conversion rate tells you that something is wrong. A funnel tells you where. Those are different pieces of information, and only the second one leads to work you can actually do this week.
The breakdown is often where the real finding sits. A checkout that looks acceptable overall can be losing almost everyone on mobile while performing well on desktop, and in a mobile-heavy market like Nepal that is the norm rather than the exception. The averaged number hides it; the funnel split by device shows it in a glance.
Common mistakes with funnel exploration
The most frequent is building the funnel out of page views when the real steps are events. If your form sits on the same URL as the page before it, a page-based funnel has no way to see the form being started, and the drop it reports is imaginary.
The second is treating every drop as a failure. People leave to check a price, ask a partner or come back on a laptop, and a healthy funnel still narrows sharply. The third is comparing funnels built with different step definitions or different open and closed settings, which produces a change that exists only in the settings.
How to act on it
Build the funnel from events that mark real intent, so make sure those events exist before you start. Keep the steps few and meaningful; a long funnel is harder to read and every extra step adds noise.
Then take the largest single drop and look at it by device and by channel before changing anything, because the fix for a mobile form problem is not the fix for a mismatched ad. Where the loss is genuinely in the page rather than the tracking, the work belongs in conversion rate optimisation. If the steps you need are not being recorded at all, fix the event tracking first — a funnel built on missing data is worse than no funnel.
