The form that lost half its users

A registration form had seven fields and lost about half the people who started it. Six months and several redesigns later it had seven fields and lost a tenth.

The number of fields was never the problem. The analytics showed where people stopped, and they stopped at two places. The first was a date of birth entered through three dropdown menus, which on a phone meant nine taps and two scrolls through a list starting at 1900. The second was a password field with rules that were only revealed after a failed attempt, so a user learned about the symbol requirement by being told their chosen password was invalid.

Both fixes were small. The date became a single text input accepting several formats. The password rules appeared above the field before typing started, and the field accepted anything long. Neither change removed a question; they changed the cost of answering one.

The lesson the team drew was about where to look. A form is not experienced as a list of questions but as a sequence of small decisions, each with its own cost, and the total abandonment is set by the two or three worst moments rather than by the length. Shortening the form by removing a cheap field improves nothing; fixing an expensive field improves everything after it.

The method is reusable and does not need much tooling. Record where people stop, in order. Attend to the biggest drop, not the longest step. Then measure again, because the second-biggest drop is often created by the fix for the first.