Conversion funnel: stages, examples and how to find where people drop off
A conversion funnel is the path visitors take towards one action you want, such as an order or a signup, counted as how many people reach each step. Divide each step by the one before it and the biggest drop shows you where to look first. The example below is made up; the method is yours to repeat on your own numbers.
Turn visitors and conversions into a conversion rate, then see what a lift is worth.
TL;DR
- A conversion funnel is the path visitors take to the one action you want, with the share who get through each step.
- Work out the step-to-step rate, not just the overall one. The biggest drop is where to look first.
- Change one step at a time and measure that step's rate before and after.
Stickybeak counts goals for page visits and events, with a conversion rate and top channels for each, and keeps the trail of pages read before each order. It is in early access, with a free plan and paid plans from US$20 a month.
Get early accessWhat is a conversion funnel?
Not everyone who lands on your site buys. Some leave straight away, some look at a product and go, some put it in the cart and never check out. A funnel lists those steps in order and counts the people at each one. It is called a funnel because every step has fewer people than the one before.
It answers a question the overall conversion rate can't. A 2.4% rate tells you how you are doing. A funnel tells you where the other 97.6% went.
Conversion funnel stages
The stages depend on what the visitor is there to do. Use the steps your buyers really pass through, not a textbook list of awareness, interest and decision.
| Kind of site | A typical set of steps | The conversion |
|---|---|---|
| Online shop | Visit, product view, add to cart, checkout, order | An order |
| Software or app | Visit, pricing page, signup form, account created | A signup |
| Local service | Visit, service page, contact page, enquiry sent | An enquiry |
| Newsletter or blog | Visit, article read, signup box seen, email given | A subscriber |
A conversion funnel example, with the drop-off at each step
A small online shop, one month, made-up numbers. 10,000 visits went through five steps.
| Step | People | Step rate | Share of all visits | Lost on the way here |
|---|---|---|---|---|
| Visits | 10,000 | n/a | 100% | n/a |
| Product view | 4,000 | 40% | 40% | 6,000 |
| Add to cart | 800 | 20% | 8% | 3,200 |
| Checkout | 480 | 60% | 4.8% | 320 |
| Order | 240 | 50% | 2.4% | 240 |
Read it two ways. The step rate is people at a step divided by people at the one before: 800 / 4,000 = 20% add to cart. The overall rate is the last step over the first: 240 / 10,000 = 2.4%.
The checkout drop looks the worst at first glance, since half of the people who reach it leave. But it loses 240 people. The step from product view to cart loses 3,200. Count the people lost as well as the percentage, and start where the most are.
A quick test of where to spend effort: lift one step by a tenth of its rate and see what it does to orders. Take the cart step from 20% to 22% and carts go from 800 to 880. At 60% and 50% after that, orders go from 240 to 264. Lifting the checkout step by a tenth, from 60% to 66%, gives the same 264. A tenth more at any step is a tenth more orders, so the real question is which step is easiest to lift by a tenth, and that is usually the one with the most people leaving and the most obvious reasons.
How to build a conversion funnel from your analytics
- Pick one conversion. An order, a signup or an enquiry. One funnel per conversion; don't mix them.
- List the steps a buyer must pass through. For each, decide what counts as reaching it: a page visit (
/cart), or an event (a button click, a form sent). - Count the people who reached each step over the same dates, so the steps are comparable. Count people with people, or sessions with sessions; see unique visitors, users and sessions.
- Work out the step rate and the overall rate as in the table above. The conversion rate calculator does the division for any pair of steps.
- Split it by source and device. A funnel for all visitors averages groups that behave differently. Mobile visitors who leave at checkout and newsletter readers who sail through can hide each other. Tagged links make the source split possible: see UTM parameters explained.
Doing it in Google Analytics 4
GA4 has a ready-made report. In Explore, start a new exploration and choose the Funnel exploration template. Google describes it as a way to "visualize the steps your users take to complete a task and quickly see how well they are succeeding or failing at each step". A funnel can be open, where people can enter at any step, or closed, where they must enter at the first (Google Analytics Help, Funnel exploration, checked 29 October 2026). Pick closed if you want the drop between steps to be about the same people.
Doing it with goals in Stickybeak
Stickybeak doesn't draw an ordered funnel chart. What it has is a Goals card: you add a goal for a page visit (such as /cart) or a custom event, and each goal shows the unique visitors who completed it, its conversion rate against all visitors in the range, and the top channels those visitors came from. One goal per step, same date range, gives you the counts for the table above, and the channel column starts the source split for you. The goal counts aren't guaranteed to be the same people from step to step, so treat the result as a map of where to look, not an audit. The custom goals and goals and revenue docs have the steps.
Where visitors usually drop off, and what to check
These are the common causes at each step, not promises. Check the cause on your own pages before you change anything.
| Step | Common leaks | What to check or try |
|---|---|---|
| Visit to product view | The page didn't match what the link or search promised. Slow load on a phone. Wrong visitors. | Split by source: one channel far worse than the rest points at a mismatch. Open the page on mobile data. See bounce rate. |
| Product view to cart | Price not clear. No sense of what you get. Doubts about delivery or returns. The button is hard to find. | Put price, delivery and returns where the button is. Add the detail a buyer would ask about. Make the button obvious on a phone. |
| Cart to checkout | Surprise costs such as shipping appearing late. A forced account. Cart that doesn't save. | Show the full cost before the cart. Allow guest checkout. Test on a phone. |
| Checkout to order | Long forms, few payment options, errors on submit, a payment page that looks different from the shop. | Walk through it yourself on a phone, with a real card if you can. Watch for error messages and required fields that aren't needed. |
A drop at one step can also be a measurement problem. If a step is a page visit and the step before it is an event, you are comparing different things; a checkout that happens on another domain can look like everyone vanished. Check the count by hand with one test visit before you trust the number.
How to measure whether a fix worked
- Write down the step rate before. The one step you will change, over a stretch of at least a few weeks of traffic.
- Change one thing. If you change the price, the button and the photos together you won't know which did it.
- Keep the traffic the same. A big email send or a new ad in the middle of the test moves the mix of visitors and so the rate. Compare like with like: the same channel, the same days of the week.
- Wait for enough people. A step with a few dozen visitors swings a lot from one more or one fewer. The same warning as in bounce rate applies: ten visits and one more changes the number by points.
- Read the step rate, then the overall rate. A better step rate that leaves orders flat usually means the people you pushed through are less likely to buy at the next step. Look at the next step too.
For sites that sell something, the other half of the picture is where the people who did buy came from and what they read first. Stickybeak keeps that trail on each order: the entry page, the referrer, every UTM tag and the pages read before buying.
Questions
What is a conversion funnel?
The series of steps a visitor takes towards one action you want, such as an order or a signup, counted as how many people reach each step. It is called a funnel because fewer people reach each step than the one before.
What are the stages of a conversion funnel?
They depend on what you sell. A shop usually has visit, product view, add to cart, checkout and order. A service site might have visit, pricing page, contact form and booking. Pick the steps a buyer really passes through, not a textbook list.
How do you calculate funnel conversion rate?
Divide the people at a step by the people at the step before it for the step rate. Divide the people at the last step by the people at the first for the overall rate. In the worked example, 240 orders from 10,000 visits is 2.4% overall.
What is a good conversion rate at each step?
There is no number that fits every site, because products, prices and traffic sources differ. Compare each step with its own past rate and with the same step for a different channel. A step that falls after a change is a better lead than any average.
How do I set up a funnel in Google Analytics 4?
In Explore, start a new exploration and choose the Funnel exploration template. Google describes it as a way to visualize the steps users take to complete a task and see how well they are succeeding or failing at each step. Funnels can be open, where users can enter at any step, or closed, where they must enter at the first step.
Can I build a funnel without a funnel report?
Yes. Count the visitors who reached each step over the same dates, using a page-visit goal or an event for each, then divide step by step on a calculator. It is less tidy than an ordered report, because the counts are not guaranteed to be the same people, but it finds the biggest leak.