A completion rate tells you how many people finished your survey. It doesn’t tell you why the rest didn’t. Somewhere between opening the survey and the submit button there is a question that made them leave, and averages won’t point at it. The funnel tab will.
[Image: The Funnel tab showing the stage list for a survey: a Start row with the total count, one row per question with count and percentage remaining, drop percentages between stages, time-per-stage badges, and the “Biggest drop” marker on one stage.]
Reading the funnel
The funnel lists your survey as respondents experienced it: a Start row with everyone who opened it, then one row per step in order. Each step shows how many people reached it, what share of the original audience that is, and the drop from the previous step - as a percentage and as an absolute count, because “-12%” reads differently when it means four hundred people. The steepest fall is flagged as the biggest drop, so the problem question is highlighted rather than hunted for.
Partial responses aren’t an option here; they’re the subject. A funnel is made of the people who didn’t finish, which is why this tab has no partial-responses toggle - it always counts everyone who started.
Time spent tells you why
Next to each step sits the time respondents spent on it. Most tools record only the total time from start to submit; the funnel keeps time per step, which is where the information is. Combined with the drop numbers, it separates the two ways a question fails. A step where people leave quickly usually asks something they don’t want to answer - too personal, too early, too presumptuous. A step where people spend a long time and then leave asks something they couldn’t answer - unclear wording, too many options, a matrix that doesn’t fit on a phone screen. The fix is different in each case, and the funnel shows you which one you’re looking at.
Long time without a drop is worth a glance too. People are tolerating the question, but tolerance is a budget; an expensive question early in the survey is paid for by drop-offs later.
[Image: Close-up of two funnel stages side by side: one with a short time badge and a steep drop, one with a long time badge and a steep drop, illustrating the “won’t answer” versus “can’t answer” patterns.]
Compare channels and time windows
Two filters frame the funnel. The channel filter shows one distribution channel at a time - the survey behaves differently when it arrives by email to your customers than when a stranger scans a QR code on a poster, and their funnels shouldn’t be averaged together. The time window (all time, last month, last week) is how you check your own work: fix the problem question, then watch whether last week’s funnel actually improved on the one that made you edit.
From finding to fix
The funnel is the diagnostic half of an editing loop. The usual repairs are unglamorous: cut the question that bleeds respondents, move the demanding one later so it’s answered by people already invested, split a heavy page in two, or make an optional question actually optional. Edit, share, and read next week’s funnel to see if it worked.
Drop-off analysis is often the feature that pushes a survey tool into its business tier. Here it’s a tab, on every plan, for every survey. Completion rate itself is measured automatically too - it can sit on your dashboard as a KPI with its own trend and health bands. And once people do finish, the results overview takes over: per-question charts, filters, and segment analysis on everything they answered.