Insights / Guide / Conversion rate optimisation (CRO)
GuideHow to read your checkout funnel
A funnel report shows where visitors leave your site. Reading it well, step by step and segment by segment, points to the fixes with the most value.
Every online business has a conversion rate, and most leaders know theirs. Fewer know where, between arriving on the site and paying, the rest of their visitors go. That is what a funnel report shows. Read well, it is the most useful single view of an ecommerce site’s health, because it points directly to where improvement will pay.
Read badly, it encourages teams to fix whatever looks worst on a chart, which is often the wrong thing. This guide explains how to build a reliable funnel, find the step that matters most and break it down so you understand why people leave.
Step 1: Count each stage
A funnel is a sequence of steps that most buyers pass through. For a typical online shop, five stages work well:
- Sessions: visits to the site.
- Product views: sessions that view at least one product page.
- Add to basket: sessions that add at least one item.
- Checkout start: sessions that begin checkout.
- Purchase: sessions that complete payment.
For sites with a longer checkout, split the checkout stage further: delivery details, delivery method, payment details and confirmation. Each extra step gives you more precision about where people leave.
Use sessions or users consistently throughout, and choose a period long enough to smooth out daily noise, usually four weeks. Avoid comparing periods that include a major promotion with periods that do not, as promotions change behaviour at every stage.
Check the tracking first
Before drawing conclusions, test that each step is tracked correctly on every device and browser. A tag that fails on one payment method or one browser will look exactly like a customer problem. Place a few test orders and confirm each appears at every stage.
Step 2: Find the biggest drop
Calculate the proportion of visitors who move from each stage to the next. Then look for the step that loses the most people. Two views help here.
The first is the percentage drop. It is normal for the largest percentage loss to be between sessions and product views, because many visitors arrive to browse, check a price or find a store address. A large drop there is not necessarily a problem.
The second, more useful view is value. Estimate what recovering a small share of each drop would be worth. A visitor who abandons at payment has shown much stronger intent than one who leaves the home page. Recovering one in twenty of them may be worth more than recovering one in a hundred visitors at the top of the funnel, and is usually easier.
As an illustration, consider a site with 100,000 monthly sessions. If 8,000 start checkout and 4,000 complete payment, half of all checkout starters leave. Those 4,000 people have chosen products, entered the checkout and then stopped. Improving that step by even a modest amount is often the single highest-value change available.
Step 3: Segment the funnel
An overall funnel hides important differences. Break it down by the segments most likely to behave differently, and compare each step side by side.
Device
Mobile visitors often convert at a lower rate than desktop visitors. Some of that is natural, as people browse on phones and buy later on a laptop. But a mobile checkout drop much larger than desktop’s usually signals a usability problem: small form fields, awkward address entry, payment pages that do not fit the screen or a missing mobile wallet option.
New and returning visitors
Returning visitors usually convert better. If new visitors abandon disproportionately at checkout, look for barriers to first-time buyers: forced account creation, unclear delivery costs or a lack of reassurance about returns and security.
Traffic source
Visitors from paid search, social media, email and organic search arrive with different intentions. A paid campaign that brings many visitors who never view a product may be targeting the wrong audience. Email traffic that adds to basket but does not buy may be responding to an offer that is not clearly applied at checkout.
Further segments worth checking include country, payment method and basket value. A sharp drop for one payment method, or for international customers, often points to a specific technical or pricing issue.
Step 4: Understand why
The funnel shows where people leave, not why. To understand why, combine the numbers with direct evidence.
- Session recordings and heatmaps show where visitors hesitate, scroll back or rage-click.
- Form analytics show which fields cause the most errors or abandonment.
- Short exit surveys at checkout ask one question: what stopped you today?
- Customer service logs reveal recurring checkout complaints.
- Watching five or six real customers attempt a purchase reveals problems no report will.
Look for patterns that match the segment differences you found. If the mobile checkout drop is high and recordings show people struggling with the address form, you have a clear, testable hypothesis.
Step 5: Turn findings into tests
Write each finding as a hypothesis: “Because mobile users struggle to enter addresses, adding postcode lookup will increase mobile checkout completion.” Prioritise hypotheses by the value of the step they affect, the strength of the evidence and the effort to test.
Where traffic allows, run controlled A/B tests and judge them on completed purchases, not clicks. Where it does not, make the change, monitor the funnel closely for several weeks and compare with the previous period, accounting for seasonality.
Common misreadings to avoid
Funnel reports are easy to misread, and a few mistakes recur often enough to be worth naming. The first is treating every drop as a problem. Some visitors add items to the basket simply to see the delivery cost or to save a list for later. A basket stage that loses many visitors may reflect browsing behaviour rather than a broken page. Look at what those visitors do next, and whether they return within a few days to buy, before concluding that something is wrong.
The second is ignoring visitors who buy across devices or sessions. Someone who adds to basket on a phone at lunchtime and buys on a laptop in the evening appears as an abandonment in one session and a purchase in another. If your analytics can link signed-in users across devices, use that view alongside the session funnel.
The third is reacting to small numbers. A segment with a few hundred sessions a month will show large swings from week to week that mean very little. Set a minimum volume below which you do not draw conclusions, and combine small segments or extend the period until the pattern is stable.
Finally, be careful about changes that move a problem rather than solve it. Removing a delivery cost from the basket page may improve the basket-to-checkout rate while increasing abandonment at payment, where the cost then appears as a surprise. Always judge a change by its effect on completed purchases, not on the single step you set out to improve.
Making it routine
Review the funnel monthly, by segment, with the people who can act on it: ecommerce, marketing and technology. Note any sudden change and investigate it within days, because a new drop often means something broke. Over time, a regular funnel review becomes the backbone of a conversion programme, keeping effort focused where it earns the most.
Source: Intelliversal analysis.
