# Measure whether people recover from a form error

Completion rate is useful, but it hides where a form excludes people. Observe correction, retained work and confirmed outcomes across the supported interaction methods.

By Cobnex editorial. Published 2026-09-10. Updated 2026-09-11.

## Define a completed business task

A successful form journey ends with the intended result, such as a created request or saved application. A click on Submit is not completion, and a client-side success animation is not evidence of server acceptance.

Track the stages separately: started, validation attempted, accepted and confirmed to the user. This helps distinguish confusing questions from service failures or a broken confirmation screen.

Avoid collecting sensitive field contents simply to measure the funnel. Error codes, field identifiers and state transitions usually provide more appropriate evidence.

## Examine recovery rather than just error frequency

Measure whether users correct an error and continue, how often the same error repeats and whether they abandon after losing input. An initially high error rate may be tolerable if instructions make recovery easy, while one unrecoverable error can block the task entirely.

In a synthetic study, eight people encounter a date-format error. Seven correct it after reading a specific example, while one cannot locate the message using a screen reader. The average recovery time should not conceal that blocked journey.

Investigate the mechanism with task observation. Telemetry alone cannot explain whether a person stopped because of inaccessible focus, unclear language or an unrelated interruption.

## Use a supported interaction matrix

Include keyboard-only use, representative screen-reader combinations, narrow screens and text enlargement in acceptance evidence. Record versions and the tested task so the result is reproducible.

Automated accessibility checks are one input. They do not establish that questions are understandable or that a dynamic error sequence is usable.

Where formal conformance is required, assess the applicable criteria and scope explicitly. A small set of successful usability sessions is not a substitute for that assessment.

## Keep performance and data preservation visible

Measure response delays and whether entered work survives validation and network errors. A user who must retype a long description after a timeout experiences a failure even if the second request succeeds.

Set acceptance around concrete outcomes: errors can be found, corrected values remain, required controls are operable and the final result is clear. Keep unresolved barriers tied to an owner and a reproducing journey rather than averaging them into a general quality score.

## Sources

- [W3C WAI: forms tutorial](https://www.w3.org/WAI/tutorials/forms/)
- [W3C WAI: user notifications](https://www.w3.org/WAI/tutorials/forms/notifications/)