Jeshwanth Srinivasan

Accessibility · HID Global

Scaling Accessibility at HID

European Accessibility Act readiness across HID's consumer apps.

Accessibility was left to each designer's judgment, and the European Accessibility Act deadline was coming. I set up an audit SOP, piloted it on one app, and it spread across HID's consumer products.

Role
Set up the audit SOP, ran the pilot, and moved repeated fixes into the pattern library
Team
Design and engineering, across product teams
Year
2022-2026
Platforms
Windows, macOS, iOS, Android
Methods
Flow mapping, risk rating, NVDA, VoiceOver, keyboard and focus testing, issue logging
Outcome
One SOP, adopted across HID's consumer-facing products

Audit maps and screens are from the HID Approve pilot. Internal service names are left out.

Why this case matters

This isn't one project. It's the process I set up so every product team could check accessibility the same way. It started with one app and spread.

Problem

From June 2025, products sold in the EU must meet the European Accessibility Act. Until then, accessibility at HID was left to each designer's judgment.

That meant coverage changed with whoever was on the project. There was no shared way to test, log, or hand issues to engineering.

The stakes were commercial. A product that doesn't comply can be pulled from the EU market.

What goes wrong without it

A core task ships that nobody can finish with a screen reader, and nobody finds out. A fix lands on one product and never reaches the next.

Research

I piloted the SOP on HID Approve, a consumer app on four platforms. Consumer products are what the act targets, so a consumer app was the right first test.

How I learned

  • Screen reader testing

    NVDA on Windows, VoiceOver on macOS and iOS, and the Android screen reader.

  • Keyboard and focus

    Every core task by keyboard alone, checking focus order and traps.

  • Shared audit log

    Each issue logged with severity, WCAG criterion, steps to reproduce, and a recommended fix.

What I found

  1. Finding 01

    Core interactions failed in VoiceOver.

    What I saw
    Approve and Decline, the app's core action, are swipe gestures. Both failed on iOS and Android. Many other core interactions couldn't be used in VoiceOver mode.
    What it changed
    They were fixed. It also set the SOP's first rule: audit the core task end to end, with a screen reader on.
  2. Finding 02

    Judgment left gaps.

    What I saw
    Each designer checked what they thought to check. Coverage changed from product to product.
    What it changed
    One SOP: the same platforms, tools, and log format on every product.
  3. Finding 03

    The same issues kept repeating.

    What I saw
    Contrast failures showed up in all seven flows. Text that didn't scale showed up in five. Controls with no name showed up in four.
    What it changed
    Repeated fixes went into the pattern library, so each one is solved once.

The pilot audit

I mapped seven core flows in HID Approve, on all four platforms. Each map marks the step where an issue breaks the flow, with its WCAG criterion and platform. The pilot found 32 issues. Four flows were rated high risk.

Accessibility risk and WCAG issue count for each audited flow
FlowRiskWCAG issues
Push approvalHigh risk4
Add serviceMedium risk5
Secure codeHigh risk5
Challenge-responseHigh risk5
Transaction signingHigh risk6
Service menuLow risk5
About and legalLow risk2
Total32

Audit maps

Select a map to open it at full size.

Push approval is the app's core action. Approve and Decline are gestures, and both failed on iOS and Android (WCAG 2.5.1).
Push approval High riskPush approval is the app's core action. Approve and Decline are gestures, and both failed on iOS and Android (WCAG 2.5.1).
Add service. The back button had no label, and Android never announced the loading status.
Add service Medium riskAdd service. The back button had no label, and Android never announced the loading status.
Secure code. Three buttons had no names, and keyboard users on Windows couldn't reach the screen.
Secure code High riskSecure code. Three buttons had no names, and keyboard users on Windows couldn't reach the screen.
Challenge-response. Windows keyboard users couldn't reach the screen. Android read instructions after the field.
Challenge-response High riskChallenge-response. Windows keyboard users couldn't reach the screen. Android read instructions after the field.
Transaction signing had the most issues. The Account field had no label for VoiceOver on macOS.
Transaction signing High riskTransaction signing had the most issues. The Account field had no label for VoiceOver on macOS.
Service menu. Low risk, but the delete prompt never told VoiceOver users which service they were deleting.
Service menu Low riskService menu. Low risk, but the delete prompt never told VoiceOver users which service they were deleting.
About and legal. Two issues: label contrast and text that didn't scale.
About and legal Low riskAbout and legal. Two issues: label contrast and text that didn't scale.

Screens from the audit

The screens behind the maps, as audited.

  • Approve and Decline are swipe gestures. Neither worked on iOS or Android (WCAG 2.5.1).
    Approve and Decline are swipe gestures. Neither worked on iOS or Android (WCAG 2.5.1).
  • The key, menu, and back buttons had no accessible names (WCAG 4.1.2).
    The key, menu, and back buttons had no accessible names (WCAG 4.1.2).
  • Placeholder text failed contrast, and Android read the instructions after the field.
    Placeholder text failed contrast, and Android read the instructions after the field.
  • Placeholder text on every transaction field failed contrast (WCAG 1.4.3).
    Placeholder text on every transaction field failed contrast (WCAG 1.4.3).
  • Form labels didn't scale with the system text size (WCAG 1.4.4).
    Form labels didn't scale with the system text size (WCAG 1.4.4).

Process

  1. 01Set up

    • Platforms: Windows, macOS, iOS, Android
    • Tools: NVDA, VoiceOver, keyboard only
    • A flow map per core task, marking where each issue breaks it
    • A risk rating per flow: high, medium, or low
    • An audit spreadsheet with severity, WCAG mapping, and recommendations

    Output. An SOP any designer could run the same way. The risk rating gave product and engineering one order of work: high-risk flows first.

  2. 02Pilot

    • HID Approve, because it is consumer facing
    • Seven core flows mapped and audited on all four platforms
    • 32 issues logged with severity and WCAG criterion

    Output. Core interactions that failed in VoiceOver came to light and were fixed.

  3. 03Roll out

    • Mobile Access
    • LinQ Mobile
    • Then most of HID's consumer-facing products

    Output. Each product was audited with the same SOP and logged in the same format, so results compare across products.

  4. 04Make it stick

    • Repeated fixes moved into the pattern library
    • New screens use those patterns instead of one-off fixes

    Output. A fix made once carries to the next product.

Method

The SOP audits tasks, not components. The trade-off: fewer screens per round, but the issues that block real work surface first. Every issue gets a severity, a WCAG criterion, steps, and a recommended fix. Engineering can prioritize and reproduce it without a meeting.

Patterns

Four recurring problems, and how each one was fixed.

  1. Focus order and focus traps

    After

    The confirm dialog now holds focus. When it closes, focus returns to the button that opened it.

  2. Controls without a name

    After

    Icon-only buttons now have names that say what they act on.

  3. Status shown by color only

    After

    Each status now has a text label. Color only backs it up.

  4. Unannounced errors and hard timeouts

    After

    Errors now appear next to the field and are announced. Timeouts offer more time first.

Design system

Repeated fixes went into the company pattern library. New screens use those patterns instead of one-off fixes.

Principles I used

  • Match between the system and the real world

    Audits followed real tasks, not isolated components.

  • Consistency and standards

    Dialog focus, field errors, and status labels moved into the pattern library, so every screen behaves the same.

  • Visibility of system status

    Status is written as text. Color only backs it up.

  • Help users recover from errors

    Errors appear on the field and are announced. Timeouts offer more time before the session ends.

Audit setup

  • Platforms

    Windows, macOS, iOS, Android.

  • Assistive tech

    NVDA, VoiceOver, keyboard only.

  • Issue format

    Severity, WCAG criterion, steps to reproduce, and a suggested fix.

Outcome

What changed
The HID Approve pilot logged 32 issues across seven flows, each mapped to a WCAG criterion. Core interactions that failed in VoiceOver were found and fixed.
The larger effect
The SOP spread to Mobile Access, LinQ Mobile, and most consumer-facing products. Accessibility no longer depends on one designer's judgment.
What I delivered
The audit SOP and its spreadsheet, the HID Approve pilot, and the repeated fixes in the pattern library.
What I'd do next
Retest the four patterns in the next release. A pattern that breaks in the library costs more than a one-off bug.