Build in Public

Share SaaS Accessibility Fixes While Building in Public

Explain a SaaS accessibility improvement with its checked interaction, input method, and environment. Keep wider evaluation and untested behavior clear.

By Uriel Bitton · · 9 min read

Illustrated cover for “Share SaaS Accessibility Fixes While Building in Public”.

The short answer

Share a SaaS accessibility fix by naming the barrier, changed interaction, completed checks, tested input method, and environment. Keep untested behavior visible. A component improvement can be useful progress without establishing full workflow accessibility, screen-reader experience, or WCAG conformance.

Share a SaaS accessibility fix by naming the barrier, the changed interaction, and the checks you actually completed. Keep the tested input method and environment beside the result. A useful public update can celebrate one improvement while leaving other interactions, assistive technologies, and wider accessibility evaluation clearly unresolved.

Building in public makes the distinction matter. A short release post may become the explanation a prospective user finds when deciding whether to try your product. Give that reader a concrete account of what changed rather than a broad label they must interpret for themselves.

Start with the barrier in the user’s task

Describe what someone was trying to do and where the interaction became difficult. “Keyboard focus did not return to the Filter button after closing the dialog” gives a reader more useful information than “accessibility improved.” Explain technical terms where needed: keyboard focus is the element currently targeted by keyboard input.

Then state the changed behavior. Keep the description close to the affected action, such as closing a dialog, reaching a control, or understanding an error. The smaller and more concrete the claim, the easier it is to connect it to the evidence you retained.

Separate the product change from the check. Code intended to restore focus is an implementation decision. Observing focus return during a recorded check is evidence about that behavior in those conditions. A planned review is a future commitment. Each can belong in a progress update when its status is clear.

Avoid inferring the user’s whole experience from the changed component. Someone may still need to open the dialog, understand its options, submit a change, and recover from an error. Your post should help readers locate the improvement within that larger task.

Retain the conditions behind the public sentence

Keep a private record of the product version, affected page or flow, steps, expected behavior, observed result, and date. Record the browser and operating-system versions, input method, and any assistive technology used. These details let you tell what was checked when another environment produces a different result.

Distinguish a keyboard-only check from a check with a screen reader. If no screen reader was used, the keyboard observation does not describe screen-reader announcements. If one browser and assistive-technology combination was examined, preserve that scope rather than treating it as every supported setup.

W3C’s accessibility evaluation overview explains that tools can help, but no tool alone determines whether a site meets accessibility standards; knowledgeable human evaluation is required. A public screenshot of a clean scan should therefore describe the scan and its coverage, without becoming a whole-product accessibility conclusion.

Use different labels for different evidence. An automated finding, a manual interaction check, and a session with a participant each tell you something specific. Describe what happened in each, who performed the relevant review when appropriate, and what remains unexamined.

Keep a component result separate from WCAG conformance

The Web Content Accessibility Guidelines, or WCAG, define technical accessibility requirements. WCAG 2.2’s conformance requirements apply to full pages and include complete processes. When pages form a sequence needed to accomplish an activity, the conformance requirements concern that process, not just an isolated successful step.

The publishing implication is practical: observing a dialog close correctly does not establish that the whole page or surrounding workflow conforms to WCAG. You can accurately report that interaction improvement without making the broader claim. A standards target and an established conformance result are different pieces of information.

Name the standard and version when they are relevant to real evaluation work. Explain what was evaluated and keep the supporting report connected to the claim. Do not substitute a successful demonstration, a tool score, or an implementation reference for the broader evaluation those words require.

This article concerns describing product progress. It does not evaluate any real SaaS, establish conformance, or provide a technical accessibility audit. The useful goal is a public account whose scope matches the work actually performed.

A fictional focus fix shows the boundary

Roster Fern is an invented SaaS for organizing class rosters. Its founder is revising a filter dialog before asking people to try a prototype. In the imagined earlier version, pressing Escape closes the dialog but leaves keyboard focus away from the button that opened it.

The product, versions, checks, and observations in this example are fictional. No code was changed, accessibility test performed, assistive technology used, or user session conducted for this article. The example is a way to examine publishing language, not evidence about Buildside or another product.

The imagined revised prototype returns focus to the Filter button after an Escape close. The fictional check covers that route in one desktop browser setup using keyboard input. It does not cover the separate Cancel button, screen-reader announcements, or completing a class-assignment task from start to finish.

W3C’s modal dialog pattern describes Escape closing the dialog and normally returning focus to the invoking element, with exceptions where the workflow calls for another logical destination. That is a reference for the intended interaction. Following one part of a pattern is not a complete evaluation of the dialog.

Part of the fictional work

Evidence available in the example

Public wording it supports

Closing with Escape

Focus returns to the Filter button in the imagined desktop keyboard check.

Describe that closing route and the checked setup.

Closing with Cancel

This route has not been checked in the example.

Keep its status untested; do not include it in the reported result.

Screen-reader behavior

No screen reader was used.

Make no claim about announcements or screen-reader task completion.

Completing class assignment

The full process has not been evaluated.

Report the component improvement without claiming complete workflow accessibility.

The distinction is more useful than a single “passed” label. It tells the fictional founder what she can explain now and which neighboring behavior needs attention. An untested route is an unanswered question, not an observed failure or an implied success.

Write the update around the checked interaction

Lead with the change that a reader can connect to a task. Then state the observation and its limits close enough that a short quotation keeps them together. If the change is only in a prototype, call it a prototype rather than a shipped release.

The following is proposed copy for the fictional Roster Fern example. It reports no real release or testing result:

In the revised Roster Fern prototype, closing the class filter dialog with Escape returns keyboard focus to the Filter button. The example check covers that closing route with keyboard input in one desktop browser setup. The Cancel route, screen-reader behavior, and the complete class-assignment process remain untested.

The next review will cover the other closing route and the surrounding task. This is progress on a specific interaction, not a claim that the product has completed an accessibility evaluation.

For an actual update, use your real release state and retained environment details. Describe a completed result only after checking it. If an issue was fixed locally but has not reached users, separate the development observation from availability in the public product.

Explain what someone can do with the information. A reader may want to inspect the prototype, understand a changed keyboard route, or report a problem in a supported environment. Choose the next action that fits the current product state.

Show the behavior without making the recording carry every claim

A short demonstration can make a focus change easier to understand. State the starting point, the input, and the observed destination in accompanying text. Readers should be able to understand the specific improvement without inferring it from a moving highlight alone.

Keep the shown version and relevant conditions visible in the explanation. If you edit a recording, describe any change that affects what it appears to demonstrate. A cut between opening and closing a dialog should not be presented as an uninterrupted record of the whole interaction.

Use shareable sample data and check the surrounding screen for private information. You can show a class filter with invented class names without exposing a real roster. Keep a demonstration created for public explanation separate from any fuller private evaluation record.

The guide to using a public SaaS changelog to support feature adoption covers how to connect a shipped change to the affected user task. Apply that approach after release: link the current behavior and practical instructions, while preserving the testing limits of the accessibility update.

Maintain current accessibility information beyond the progress post

A build-in-public post explains a development moment. People who encounter a barrier later need current information about the product, known limitations, and how to raise a problem. Keep that information maintained rather than requiring someone to reconstruct the situation from old founder updates.

W3C’s accessibility statement guidance recommends plain language, contact information, known limitations, and tested environments. It also distinguishes a user-facing statement from a technical assessment or declaration of conformity. A celebratory progress post should not quietly take the place of those different records.

If your product has an accessibility statement, update it when the verified facts change. Keep the statement, release note, and public explanation consistent about availability and remaining limitations. Describe an alternative or support route only when it actually exists and you can explain its limits.

A new fix can also make an older public claim inaccurate. Add appropriate dated context where readers may still rely on the old wording. Preserve the development history while giving people a dependable route to current information.

Invite feedback that helps the next review

Ask about the affected task and the behavior still unresolved. A reader may identify a different closing route, an environment you did not cover, or a point where the surrounding process becomes difficult. Give people a clear way to report those details without asking them to post private account information publicly.

Do not turn a general request for comments into a claim that users with disabilities evaluated the product. If you want structured evaluation with participants, arrange it explicitly and describe what participation involves. Report who actually participated and what was observed only with appropriate permission and context.

Keep a reported problem connected to the relevant version and steps. A reader’s account can guide investigation; reproducing the issue and checking a change are further actions. Publish those actions as they occur rather than compressing a report, fix, and verification into one assumed outcome.

For your next public update, choose one barrier and trace its record from the original difficulty through the changed behavior and completed checks. Write the limitation beside the improvement. Then name the next review that would let you make a more useful claim.

Questions founders ask next

Should I wait for a complete audit before sharing a small fix?

You can share a verified improvement before a wider evaluation is complete. Describe the affected interaction, actual checks, release state, and remaining scope. Keep any broader conformance statement tied to the evidence required for that separate claim.

Can an automated score establish that the fix works for screen-reader users?

An automated score reports the checks the tool performed. It does not establish a person’s experience with a screen reader or completion of the relevant task. Describe any actual assistive-technology checks or user sessions separately, with their conditions and limits.

What if a reader reports a barrier after I announce the improvement?

Record the reported task, product version, environment, and behavior through an appropriate feedback route. Investigate the difference from your earlier check. Update the public account when verified findings change it, while keeping the original scope and timeline clear.

Sources

Frequently asked questions

Should I wait for a complete audit before sharing a small fix?

You can share a verified improvement before a wider evaluation is complete. Describe the affected interaction, actual checks, release state, and remaining scope. Keep any broader conformance statement tied to the evidence required for that separate claim.

Can an automated score establish that the fix works for screen-reader users?

An automated score reports the checks the tool performed. It does not establish a person’s experience with a screen reader or completion of the relevant task. Describe any actual assistive-technology checks or user sessions separately, with their conditions and limits.

What if a reader reports a barrier after I announce the improvement?

Record the reported task, product version, environment, and behavior through an appropriate feedback route. Investigate the difference from your earlier check. Update the public account when verified findings change it, while keeping the original scope and timeline clear.

A note from Uriel Bitton

I share practical lessons from building Buildside and learning alongside SaaS founders.

Keep building with us

Get practical founder lessons and product updates in your inbox.

Subscribe or Join Buildside.

Keep reading

Browse more articles from Buildside