Counting the Unseen Minutes: A 2026 Method for Auditing Screen Reader Effort in UK Public Forms

Public forms are often judged by whether they can be completed, not by what completion costs. For screen reader users, the difference between a form that is technically operable and one that is practically efficient can be several minutes per transaction — and considerably more when an error sends the user back to the start.

That difference is the screen reader overhead: the extra elapsed time, focus changes, listening effort and error recovery work required to finish the same task. It is not captured by a simple pass or fail. It needs to be measured.

What the hidden minutes are

Overhead appears in small increments. A date field split into three controls adds two extra focus stops. An address lookup that opens a modal without moving focus adds a navigation loop. An error summary that does not receive focus adds a page re-read. None of these failures is necessarily a blocker on its own. Together they turn a two-minute task into a six-minute task and make the user less confident that the form has been completed correctly.

The comparison is not between screen reader users and an abstract ideal. It is between screen reader users and sighted keyboard-only users completing the same task on the same form. Both groups may face poor labelling or confusing validation. The overhead is the additional cost imposed by the interaction pattern.

A lightweight measurement method

Product teams do not need a large research programme to find this. A small, well-controlled pattern review is enough to expose repeatable friction.

Select three representative form patterns: date entry, address lookup and error recovery. Recruit six to eight participants — three or four screen reader users and three or four sighted keyboard-only users. A sample of this size will not support statistical claims, but it will reveal whether a pattern creates a consistent penalty.

Run the sessions on the same device, browser and connection where possible. Randomise task order. Use the same task instructions and allow no assistance beyond a scripted prompt. Stop after a fixed time or after a set number of failed attempts. Record screen and keyboard activity with consent, and keep all test data anonymous.

Measure four things:

• Elapsed time from first focus on the form to the success confirmation.
• Focus changes and keystrokes required to complete the task.
• Error cycles: the number of times an error is presented and the time taken to recover.
• Post-task effort rating, using a simple one-to-seven scale.

Calculate an overhead ratio: median screen reader completion time divided by median sighted keyboard completion time. Calculate focus overhead as the difference in median focus changes. Report the figures by pattern, not as a single accessibility score. A single score hides the specific design decisions that caused the delay.

Illustrative findings across three patterns

The following pattern-level observations are illustrative of what this method tends to surface. They are not a statistical study, but they show where teams should look first.

Date entry. Three separate day, month and year combo boxes are a common UK public-service pattern. Screen reader users often have to listen to three labels, navigate three controls and manage auto-advance behaviour that moves focus without warning. The overhead comes less from typing than from orientation. A single text input with a clear label, an example format and tolerant parsing removes most of it. If segmented controls are retained, group them under one label, avoid auto-advance and keep the hint associated with the group.

Address lookup. Postcode search can be efficient, but it often opens a modal that does not receive focus, presents a results list without announcing the number of matches, and hides manual entry. Users may tab through the page behind the modal or restart the lookup. Keep manual address entry visible, move focus to the results, announce the result count, label each result with the full address, and avoid trapping focus. Allow a building number or street name to narrow the search.

Error recovery. An error summary at the top of the page is good practice, but only if focus moves to it and each error links to the relevant field. A common failure is a summary that appears visually while focus remains on the submit button at the bottom. The user hears nothing, tabs forward into the next page or re-reads the entire form. Inline errors that are visually adjacent but not programmatically associated with the field create a second pass. Move focus to the summary, link each error, preserve entered data, associate inline messages with their fields, and announce submission status in a live region.

Fixes that pay back minutes

The highest-return fixes are usually unglamorous. Use native controls where possible. Give every field a persistent visible label. Keep hints and errors programmatically associated with the field they describe. Do not move focus unless the user's action requires it. If a custom widget is unavoidable, follow an established accessible pattern rather than inventing one.

For multi-page forms, show progress in text and in the page title. Preserve data when the user moves back. Avoid timeouts that erase work. On submission, announce success or failure immediately and keep the next step close to the confirmation.

These changes reduce effort for everyone. Sighted keyboard users also benefit from fewer focus stops and clearer errors. The point of measuring overhead is not to create a separate accessibility budget. It is to show that accessibility quality has a time cost, and that the cost is design-led.

Reporting the overhead without overclaiming

Present the results as pattern-level evidence. State the sample size, the tasks, the devices and the limits of the method. Do not claim that a small study proves a population-wide average. Do claim that a consistent pattern — such as a modal that loses focus or a date field that requires three separate announcements — is a design risk worth fixing.

Repeat the measurement after the fix. If the overhead ratio falls and the error cycles drop, the team has evidence that the redesign worked. That is more useful than another checklist that says the form is compliant but says nothing about how long it takes to use.