Funnel analysis is a method of breaking a goal, such as a purchase or an enquiry, into the steps people take to reach it, then measuring how many continue from each step to the next and where the rest drop out. It shows not just how many people converted, but where the others were lost.
How funnel analysis works
You start by writing down the steps in order. For an online shop that might be: view a product, add to basket, begin checkout, enter delivery details, enter payment, purchase. For a service business it might be: land on a service page, start the enquiry form, submit it, become a qualified lead, book, pay. The last few of those happen in your CRM, not on the site, and a complete funnel includes them.
For each step you count how many people reached it and calculate two rates: the step rate, from one step to the next, and the overall rate, from the first step to the goal. The step with the largest number of people lost is usually where to look first, though a small percentage loss at a high-volume step can outweigh a dramatic percentage at a quiet one.
Some choices change the numbers:
- A closed funnel counts only people who entered at step one; an open funnel lets people join at any step.
- A time limit decides whether someone who returns a week later still counts as completing.
- Breaking the funnel down by device, channel or new and returning visitors often explains a drop that the total hides.
In GA4 the main tool is the funnel exploration, and ecommerce properties also get prebuilt purchase and checkout journey reports. CRM pipeline reports cover the steps after the enquiry.
Why it matters
Most businesses try to grow by buying more traffic. Funnel analysis often shows that the cheaper win is further down: a delivery cost that appears only at checkout, a booking calendar that fails on mobile, or a payment step that drops people at the bank’s security check. Strong Customer Authentication, which UK card payments have required since 2022, is a common place for checkouts to lose people if the payment provider handles it clumsily.
It also stops arguments being settled by opinion. Seeing that most losses happen between basket and checkout, not on the product page, points the work, and any conversion rate optimisation effort, at the right place.
Common mistakes
- Defining steps that are not really sequential, so people appear to skip or reverse them.
- Reading a tracking gap as a drop-off, for example a checkout event that does not fire on one browser.
- Judging a step on a handful of users, where the sample size is too small to mean anything.
- Stopping the funnel at the form, so you optimise for enquiries that never become customers.
- Comparing periods with different consent rates or tracking set-ups as if like for like.
How to act on it
Write your funnel on one page, from first meaningful visit to payment. Check every step is tracked by triggering it yourself and confirming it appears. Build the funnel, find the largest absolute drop-off, then look at that step in detail with recordings, error logs and a test purchase or enquiry on a phone.
Change one thing at a time and re-measure over a comparable period. When I run performance marketing, the funnel review comes before any budget increase, because sending more paid traffic into a leaking funnel only makes the leak more expensive.