Workflow drop-off analysis

Workflow drop-off analysis is the practice of measuring how many users abandon a multi-step process at each step, so you can pinpoint exactly where a flow loses people. A workflow might be a checkout, a signup, an onboarding sequence, a loan application, or an internal approval chain. At every step, some share of users who arrived do not advance. That share is the drop-off rate for that step.

The analysis matters because a single completion figure hides the story. If only 30 out of every 100 users finish a signup, the headline is a 70 percent loss. But that 70 percent is not spread evenly. It might be that 50 of those 70 abandon on the email verification step alone. Knowing that one step is responsible for the majority of the loss changes where you spend effort entirely.

Drop-off analysis is closely related to funnel and conversion rate work, but it is more diagnostic. A conversion rate tells you the outcome. Drop-off analysis tells you the location of the problem, step by step, so the next action is obvious rather than a guess.

The core calculation is simple at each step: take the users who entered the step, subtract the users who advanced, and divide by the users who entered. Doing this for every step in sequence produces the full drop-off profile of the workflow. The inputs you need are consistent across any flow.

1. 1

   Step definition

   A clear, ordered list of the discrete steps in the workflow. Each step must have an unambiguous entry event and an unambiguous completion event, otherwise the counts will not line up.

2. 2

   Users entering the step

   The count of unique users who reached the start of the step in the period. For the first step this is everyone who began the workflow. For later steps it is whoever survived the steps before.

3. 3

   Users completing the step

   The count of users who advanced from the step to the next one. The gap between entering and completing is the absolute number lost at that step.

4. 4

   Time window

   The period over which you count, and a sensible cut-off for how long a user has to complete a step before they are treated as dropped. Without a window, slow but eventual completers get miscounted as losses.

Two views matter. The per-step drop-off rate shows where the leak is. The cumulative completion rate, the share of original starters who reach the end, shows the cost of all the leaks combined. A step can have a modest drop-off rate of 10 percent yet still be the largest absolute loss in the flow if almost everyone reaches it. Always look at both the rate and the raw count, because the step worth fixing is the one losing the most actual people.

A metric tree turns drop-off analysis from a chart you read into a structure you can act on. The root is the overall completion rate of the workflow. Beneath it sit the individual steps, and beneath each step sit the reasons users abandon there. This is what makes the decomposition causal rather than descriptive.

The first level is the steps themselves, ordered as the user experiences them. Each step node carries its own drop-off rate and the count of users lost. The second level explains the loss. A step might leak because of slow load times, a confusing field, an unexpected cost, a required action the user cannot complete, or a technical error. Each of those is a separate branch with its own owner.

This is where KPI Tree changes the work. Every node carries RACI ownership, so the email verification step is owned by the team that can actually change it, not the team that happens to report the number. When the drop-off rate on a step moves, the accountable owner is notified directly rather than the regression sitting unseen in a dashboard. The gap between seeing a flow leak and fixing it is exactly the gap KPI Tree is built to close.

Drop-off varies enormously by workflow type, audience, and how much the user wants the outcome. The figures below are typical per-step ranges for common flows. Treat them as orientation, not targets. A high-intent workflow such as a paid checkout should leak far less than a cold, top-of-funnel signup where the user has little invested yet.

A useful rule is that no single step should be losing more than roughly a third of the users who reach it without a clear, accepted reason. Mandatory identity checks and payment steps will always shed some users, and that is expected. A drop-off spike on a step that used to be quiet is a stronger signal than any absolute benchmark, because it points to a recent change you can usually trace and reverse.

Improving a workflow is not about reducing every drop-off at once. It is about finding the single step responsible for the most lost users and fixing the specific cause behind it. Work the largest leak first, measure the effect, then move to the next.

1. 1

   Measuring every drop against the first step

   Comparing each step to the original starters makes late steps look catastrophic and hides where the real leak is. Always measure a step against the users who actually entered that step.

2. 2

   Ignoring slow completers

   Users who finish a step a day later are not drop-offs. Without a sensible completion window, delayed but successful users are counted as lost and the drop-off rate is overstated.

3. 3

   Chasing the highest percentage instead of the highest count

   A tiny step with a 60 percent drop-off can matter less than a busy step with a 20 percent drop-off. Optimising the eye-catching rate while ignoring the larger absolute loss wastes effort.

4. 4

   No owner on the leaking step

   A drop-off chart with no one accountable for each step becomes a recurring discussion that never resolves. Without a named owner per step, the leak is observed every week and fixed by no one.