Shaping Accessibility Across Products

How accessibility gets built into products across Meta.

Timeline
December 2023–July 2026
Role
Product Designer · Product Systems · Interaction Design · AI Strategy
Team
Engineering · Product Management · Product Design · Content Design · Research · Marketing · Learning Program · Legal

Accessibility had relied on specialists catching issues late in development. The European Accessibility Act made it business-critical, and as AI accelerated how teams built, reviewing after the fact couldn't keep up. The deeper problem was that a standard only becomes real when people can find it, apply it, verify it, and trust it—and at Meta's scale, none of that happened on its own.

Accessibility guidance organized around product decisions The Accessibility at Meta site on meta.com An accessibility knowledge base structured for retrieval

Approach

Accessibility started as something specialists upheld. I shifted more of that responsibility into the product itself: how people find support, teams evaluate quality, and guidance gets surfaced.

What changed

  • Product teams could apply the accessibility standard during design and build. 29 requirements → 23 decision-oriented topics

    The Playbook translated accessibility requirements into practical guidance organized around decisions teams were already making, with patterns, examples, and implementation guidance they could use during design and build.

  • Accessibility review stopped being a manual bottleneck. 3–5 hour review eliminated · 1,711 hours saved

    Shared tooling reduced the manual coordination required across program managers, reviewers, and product teams. Automated validation eliminated a recurring 3–5 hour review per wave, while a separate submission workflow saved 1,711 hours.

  • AI could find and ground teams in the guidance that applied. 200+ pages · deprecated results 25% → 0

    Instead of searching across 200+ pages, teams could start with an accessibility question. The system interpreted what they were trying to do, routed them to the authoritative source, and surfaced guidance scoped to that task, reducing deprecated results from 25% to zero.

  • People could find accessibility support in one place for the first time. 30+ features · 45 Help Center articles

    A needs-based model became a customer-facing experience on meta.com, organizing accessibility around what people needed rather than which product they were using and creating a direct path from feature discovery to Help Center support.

    View public accessibility work

Public footprints

Full case study Go deeper into the work Need access? Request the password at ammerallj@gmail.com
Next case study Giving Components an Identity Every object in Microsoft 365 belonged to the app it lived in. This one had to belong to itself.