Find out if length was really the problem - or just the obvious guess.
The room's instinct was “just make it shorter.” My job was to check the evidence before touching a single screen.
“Before touching a single screen, I wanted to know the exact question people were rage-quitting on.”
Completion was leaking, and averages hid the story.
I needed to move from “people drop off” to “people drop off here, because of this.”
- High start rate, low completion - effort was being wasted at the worst possible moment.
- No visibility into which step or question caused abandonment.
- Assumptions (“it's too long”) competing with no evidence either way.
Numbers to find where; humans to learn why.
First, a step-by-step funnel to locate the exact drop-off cliffs. Then qualitative sessions to understand the why behind them.
Charting completion per step turned a vague “drop-off” into two specific, fixable cliffs.
The research surfaced clear culprits:
- A jarring format shift mid-survey that made people feel it was restarting.
- No sense of progress - users couldn't tell if they were near the end, so they bailed.
- A couple of heavy, open-ended questions placed exactly where energy was lowest.
- Ambiguous wording that made a few questions feel like traps.
The fix wasn't “make it shorter” - it was make it feel shorter and clearer: a visible progress indicator, smoother question grouping, lighter interactions early and heavier asks reordered.
More voices reached the end - and leadership could trust them.
By fixing the specific friction rather than blindly cutting content, the data feeding leadership got both bigger and more trustworthy.
