Most forms are abandoned at one field, and it is rarely the field anybody suspects. Form analytics exists to name it. The useful part is not that a form has a completion rate — you can get that from two pageview counts — but that the giving-up has a location.
What is collected, exactly
For each field a visitor touched: a stable selector, how many separate times it was focused, the total time it held focus, and how many edits it received. For the form as a whole: whether it was submitted, whether it was abandoned, and which field was focused last if it was.
What is not collected is anything anybody typed. Not redacted afterwards, not stored and masked — never read in the first place. Buttons, hidden inputs and reset controls are skipped as well, because a button press is a click and a hidden field has no engagement to measure.
Three shapes, three different problems
The counters matter more than the drop-off field, because they say what kind of trouble a field is causing.
- Many focus episodes, short each time. People are leaving the field and coming back — checking a reference number, re-reading a label, comparing against another field. Usually a labelling problem: the field is asking for something the visitor has to go and find.
- One long episode, few edits. Hesitation rather than difficulty. The field is asking for something people are reluctant to give, and the fix is often to explain why it is needed or to stop asking.
- Many edits. Correction, and correction almost always means validation. A field that rejects input the visitor believes is correct — a phone number with spaces, a postcode without — produces this shape reliably.
The drop-off field is a starting point, not a verdict
The field blamed for an abandonment is the last one focused before the visitor left. That is the best single guess available and it is still a guess: somebody who gives up on a long form often does so while their cursor rests in a field that gave them no trouble at all.
So read it alongside the counters. A drop-off field with one short focus and no edits is probably innocent — the decision was made earlier. A drop-off field with the highest edit count on the form is the one to fix first.
Field order is recorded, and it is more useful than it sounds
Fields are recorded in the order they were first focused, which is not necessarily the order they appear. People skip. A form where visitors consistently jump past the third field and return to it at the end is telling you that the third field is the hard one, regardless of whether anybody abandons there.
The fixes that usually work
In rough order of how often they help: remove the field entirely, because the commonest cause of abandonment is asking for something you do not need yet; explain the field, if it is asking for something whose purpose is not obvious; loosen the validation, if people are correcting themselves repeatedly; and split the form, if the drop-off is spread evenly rather than concentrated — an even spread usually means length rather than any one question.
What rarely helps is making the button bigger. If the counters say people are struggling three fields above it, the button was never the obstacle.
Form analytics is collected by the same script as click, scroll and attention maps, so there is nothing extra to install. Where a form is failing for a reason the counters cannot show, session replay is the next place to look — and worth reading first, because what it records is not what most people assume.