Glossary · Measurement
Form drop-off analysis
Seeing exactly which question people stop at — the feature even competing vendors name as the one worth paying for.
Also: Per-question drop-off, field-level analytics
One of 25 entries in the Endpoint Forms glossary.
Definition
Measuring, for every field or step, how many people reached it and how many continued past it. The output is a curve down the form, and the cliff in that curve is the thing you fix.
Why it matters
It converts an argument into an observation. Without it, deciding whether to drop the phone-number field is a debate about best practice; with it, it is a number.
“A lot of people spend tons of time optimizing the landing page and almost none looking at what happens inside the form. … Tracking drop-off per question is super valuable because it stops the guesswork. Instead of debating what might be the issue, you can see exactly where people hesitate.”
The corroboration worth more than the endorsement is that this is what vendors themselves name when asked what buyers would pay for:
“The one I’d actually pay for is drop-off reporting. Not just seeing how many people completed the form, but being able to see exactly where people stopped along the way.”
In practice
- The unit is the field, not the form. A step-level curve on a multi-step form is a start; a field-level one inside each step is the answer.
- You need partial submissions stored server-side. Client-side events alone lose the sessions that closed the tab.
- Segment by traffic source before reading it, or the paid and organic curves average into a shape neither audience has.
- Watch time-per-field alongside exit rate. A field people sit on for forty seconds and then complete is costing you conversions further down.
Conditional logic complicates the arithmetic and it is worth getting right: a question only 30% of respondents ever see has a denominator of its own. Compare like with like — see conditional logic.
The common mistake
Reading it in aggregate.
The single most common way this analysis fails is being averaged across everyone. The finding in the corpus was not “people drop off” — it was that paid traffic would not tolerate a field that organic traffic sat through without complaint.
Averaged together, that field looks acceptable. Split by source, it is an obvious change to make on the paid landing variant and no change at all elsewhere. Aggregate reporting does not just weaken the finding; it deletes it.
Related terms
- Form abandonment
- Someone started your form and left — the only funnel metric most teams infer rather than observe.
- Partial submission
- What someone typed before they left. Both a recoverable lead and the raw material for every drop-off report you will ever run.
- Completion rate
- Submissions divided by the people who could have submitted — and the denominator is the half nobody agrees on.
- Conditional logic
- Showing, hiding, skipping or routing based on answers — the category’s number one functional complaint, and the missing piece is a debugger.
- Hindsight split testsOurs
- Our name for split tests ranked on Yield rather than completion rate — where the winner cannot be known at submit time, so the test waits.