Funnel Drop-Off Calculator: Find Your Costliest Leak

Enter the users who reach each step of your funnel and this shows where they leave, what each leak is worth in orders and revenue, and which step deserves your next test. The step that sheds the most people is usually not the one costing the most money: a leak next to the checkout is worth far more per user than a leak at the top.

How the value figure is worked out

Counting the people who leave a step tells you very little on its own, because a visitor who bounces off a category page and a visitor who abandons a filled basket are worth wildly different amounts. So each leak is valued by how the survivors of that same step behave: if the people who reach your basket go on to buy a quarter of the time, then each extra person you keep there is worth a quarter of an order.

That gives a tidy piece of arithmetic. The orders lost at a step come out as your finished orders multiplied by (1 divided by the step rate, minus 1). Rank leaks that way and the order never depends on how much traffic sits above the step, only on the step's own conversion rate. This is why the biggest drop-off on the chart is such a reliable trap: it usually sits at the top of the funnel, where the numbers are largest and the visitors are least committed.

Getting the inputs right

• Count unique users, not sessions or events. Someone who views four product pages is one user at that step. Mixing the two inflates the middle of the funnel and hides the leak.
• Use one date range for every step. Pulling one number from last month and another from last week produces step rates that never happened.
• Keep the steps genuinely sequential. If people can reach checkout without touching a product page, the step rate above it is meaningless. Either restrict the funnel to the path most buyers take or drop the optional step.
• Segment before you conclude. Mobile and desktop funnels usually leak in different places, and a blended funnel can hide both. Run it twice.

What the recovery figure does and does not promise

The last block assumes the users you win back behave like the users who already got through that step. In practice they tend to convert a little worse, because they had a reason to leave, so treat that number as the ceiling on a fix rather than a forecast. Its real job is to answer whether the work is worth commissioning at all: if recovering a tenth of your worst leak buys a rounding error, the leak is not where your effort should go, however ugly the percentage looks.

Once you know which step to attack, the diagnosis comes from research rather than arithmetic. Session recordings, form analytics, and exit surveys on that one page will tell you why people leave, and the funnel optimisation guide covers how to work through it. When you have a fix worth testing, size it with the sample size calculator and the duration calculator before you launch, then put a number on the win with the revenue uplift calculator. If your end-to-end figure looks low, what counts as a good conversion rate puts it in context.