How is funnel drop-off rate calculated?
The formula is straightforward:
drop_off_rate(step_n) = 1 - (users_who_reached_step_n+1 / users_who_reached_step_n)
Or equivalently:
drop_off_rate(step_n) = users_who_left_at_step_n / users_who_entered_step_n
Both forms produce the same number. The denominator is always users who entered the step, never the users who entered the funnel. That distinction is what separates per-step drop-off from overall funnel completion.
What counts as "entry" into a step?
The definition matters more than people expect. A user has "entered" a step when they have done the action that qualifies them as eligible for the next one. Common conventions:
- Pageview-based: the step's URL was loaded
- Event-based: a specific event fired (e.g.
add_to_cart,signup_started) - Element-based: a specific element became visible in the viewport
Each definition produces different numbers from the same data. Pageview-based funnels overcount entry (people who loaded the page but never interacted), while event-based funnels can undercount it if the event fires too late in the user's intent.
Worked example with real numbers
Take a four-step signup funnel:
| Step | Users entered | Users continued | Drop-off rate | Cumulative completion |
|---|---|---|---|---|
| 1. Landing | 1,000 | 600 | 40% | 100% |
| 2. Signup form | 600 | 200 | 67% | 60% |
| 3. Email verify | 200 | 50 | 75% | 20% |
| 4. First action | 50 | — | — | 5% |
Notice that step 3 has the worst per-step drop-off (75%) even though step 2 lost more raw users (400 vs 150). Per-step drop-off is what you optimize; cumulative completion is what you report to the board.
How is drop-off different from completion rate?
| Metric | Definition | Example (step 2 above) |
|---|---|---|
| Step drop-off | % who left at this step | 67% |
| Step conversion | % who advanced from this step | 33% |
| Cumulative completion | % of original cohort still in the funnel | 20% (after step 3) |
| Overall conversion | % of original cohort that finished | 5% |
They are mathematically related but answer different questions. Drop-off says where the leak is; completion says how big the bucket got.
Why isn't a bigger drop-off always worse?
A high drop-off rate is only bad if the people leaving were the right people to keep. Some funnels are deliberately designed to qualify users out:
- A pricing page between "interest" and "checkout" filters out tire-kickers
- An account creation step before a free trial filters out bots
- A credit card capture before a free trial filters out users who will never pay
In those cases, an 80% drop-off can be healthier than a 40% drop-off, because the survivors close at a much higher rate. The metric to watch alongside drop-off is the downstream conversion of survivors: if filtering is working, the next step's conversion goes up enough to compensate.
What about closed vs open funnels?
A closed funnel only counts users who entered at step 1. A user who arrives directly at step 3 (e.g. via a deep link) is excluded entirely. An open funnel counts whoever shows up at any step.
Drop-off rate behaves differently in each:
- Closed funnel: cleaner attribution, smaller volumes, false negatives for organic deep traffic
- Open funnel: larger volumes, but step entries can be inflated by users who skipped earlier steps and may not represent the intended journey
Pick one and stick to it across reports. Mixing the two is the most common reason funnel numbers stop matching across dashboards.
How it relates to CloseTrace
CloseTrace builds funnels from any combination of pageviews, clicks, and custom events, computes per-step drop-off in real time, and lets you click any drop-off bar to load every session that abandoned at exactly that step. The replay shows you the why behind the percentage instead of leaving you to guess.