Skip to main content
Meytal Dahan
Back to insights

Meeting Accessibility Standards (WCAG) in Public Systems: How Do You Design a System That Withstands Public Scrutiny?

Engineering leaders working on public systems know that meeting accessibility standards isn't a "Nice to Have" - it's a legal requirement. Any system intended for the general public must meet WCAG (Web Content Accessibility Guidelines) standards at level AA at minimum, and failing to do so can lead to lawsuits, demands for expensive remediation, and damage to the organization's public reputation. On the government project, accessibility was a first-order requirement, not a late one. We designed every UI component to meet accessibility requirements from day one: sufficient color contrast, keyboard-only navigation, alt-text descriptions for every visual element, and support for screen readers. This wasn't done in a "final stage" - it was built into the skeleton of the Design System. In addition, development received a handoff package that included not only the visual design, but also explicit guidelines for the behavior of every component - ARIA Labels, Tab order, and response to keyboard states. The development team didn't have to "invent" how to make things accessible - everything was defined in advance. For CTOs working on public systems, or on systems that require compliance with accessibility standards, the insight is this: require your product designer to understand accessibility in depth, and to deliver a handoff package that includes explicit guidelines for every component. Accessibility added at the end of a project costs 10x more than accessibility built in from the start.

Get in touch

Have a project in mind?

Drop a line. Meytalyav@gmail.com

Related articles

Government & Public Sector
FoundersFrom MVP to Full Version

Why GovTech MVPs Don't Look Like Startup MVPs

Founders are taught to ship the scrappiest MVP and iterate in public. In govtech, that instinct can sink you. A citizen service that looks like a beta — or breaks for a screen reader user — loses trust you may never win back. So I redefine 'minimum': not a thin slice of everything, but one complete, accessible, end-to-end path a real person can finish with dignity. Narrow the use cases brutally. Never narrow the quality. Then grow in concentric circles — keep the simple core intact, layer depth around it. Your launch version should be something you'd defend in a public hearing. That's the bar.

Read
Government & Public Sector
CEOsPersonalization & Tailored Models

Personalization That Survives Public Scrutiny

When CEOs hear 'personalization' they picture a consumer feed that learns you. In a public service, that's a liability — citizens didn't opt into being profiled, and any hint of differential treatment invites fairness questions you don't want to answer in public. So I reframe it as relevance, not prediction. Not a model guessing what you want — a service showing each person only what actually applies to them, driven by their situation, not surveillance. A parent sees the parent path. Someone who already applied sees status, not a fresh form. That cuts confusion and call volume, and every tailoring decision stays explainable. The efficiency of personalization, without the consumer-playbook downside.

Read
Government & Public Sector
Project ManagersUsability Testing

Usability Testing as Schedule Insurance for Public Services

Usability testing isn't a threat to your timeline — it's insurance on it. The costly defects aren't the ones found in a session; they're the ones found after launch when the fix means re-planning everyone's quarter. Test early, test on the worst device, and label every finding by whether it blocks the core task. A PMO can schedule around that. It can't schedule around surprises.

Read
Meytal Dahan

About

Making complicated into easy for users.

Senior product designer with a decade of work across complex systems - financial risk platforms, legal operations, healthcare apps, manufacturing tooling and insurance portals. The common thread is depth: products where the data is rich, the users are expert, and the interface has to disappear into the work.