Multi-Step Funnel Analysis Calculator

Multi-step funnel analysis breaks a user journey into a sequence of stages (visit, add to cart, checkout, purchase) and measures the conversion rate and drop-off between each consecutive pair, plus the overall conversion from the first stage to the last. Unlike a single conversion rate, which only compares two endpoints, a multi-step funnel shows exactly where along the path people are lost, not just how many made it through overall. This calculator supports 2 to 10 stages, so it can model anything from a simple two-step signup flow to a longer e-commerce or onboarding journey.

Step conversion is the current stage's count divided by the previous stage's count. Step drop-off is 1 minus that same ratio. Overall conversion is the final stage's count divided by the first stage's count. In the worked example, 1,800 people reaching "Add to cart" out of 10,000 who reached "Visit" gives an 18.00% step conversion and an 82.00% drop-off for that step. Each step's rate only depends on the two stages immediately next to it; it doesn't reference any other stage in the funnel, so a change further upstream or downstream doesn't directly move that one step's own number.

It usually signals a data problem: double-counting, a broader date range applied to the later stage, or two stages that were actually measuring different events despite sharing a funnel. The calculator flags this directly on the affected step rather than silently displaying a conversion rate above 100% as if it were routine. It isn't always an error, though: different cohort definitions or measurement windows between stages can occasionally produce this legitimately. Either way, the calculator computes exactly what's entered for each stage; it doesn't independently confirm that a later stage's count is truly a subset of an earlier one.

At least two, and up to ten. Two stages is functionally the same shape as a simple conversion rate calculation: one ratio between a starting count and an ending count. The value of a multi-step funnel comes from adding more stages in between, since that's what reveals which specific transition is losing the most people, rather than just the aggregate first-to-last number. There's no requirement to use every stage a real journey has; a funnel with the 3 or 4 stages that matter most for diagnosis is often more useful than every micro-step logged in the underlying data. Adding stages that never see meaningful drop-off mostly adds noise to the table without adding diagnostic value.

That step's conversion becomes undefined, since dividing by a zero-count previous stage has no meaningful result - the calculator won't display it as a real percentage. Any step downstream of a zero-count stage is also not meaningfully computable in the usual sense, since the funnel effectively broke at that earlier point. A zero count partway through a funnel is often itself the finding worth investigating - it can mean genuinely no one reached that stage, or it can mean the stage wasn't tracked for that period, which are two very different explanations worth ruling between. A zero at the very first stage is a special case too - with nothing to divide by anywhere downstream, none of the funnel's rates can be computed at all.