Inclusive design answers a different question from accessibility:
Does this respect the real-world context, identity, language, culture, device, confidence level, and constraints of the person using it?
Accessibility asks whether someone can operate the interface with their abilities and assistive tools. Inclusion asks who gets left out by the assumptions baked into the product — the person on a $40 phone over a metered 2G connection, the person whose name doesn't fit your form, the person reading in their third language, the person who has never done this online before and is afraid of getting it wrong. WCAG conformance does not address any of that. This skill does.
Per W3C, accessibility, usability, and inclusion overlap but each has a distinct focus: accessibility targets disability, usability targets the quality of the experience (effective, efficient, satisfying), and inclusion targets the full breadth of human diversity — hardware and software, literacy, economic situation, education, geography, culture, age, and language.
Use this skill when the work is about who you might be excluding and why — not about a specific assistive-tech bug. Triggers: reviewing a design "through an inclusive lens," building personas, internationalization/localization, designing name/address/profile forms for a global audience, supporting low-end devices or poor connectivity, lowering cost/data barriers, onboarding nervous or first-time users, reducing stress and cognitive load, or any "are we leaving anyone out?" conversation.
When not to use it: if the task is screen-reader support, keyboard navigation, focus order, color contrast, alt text, ARIA, captions, or passing WCAG — that's the accessibility skill. The two are partners; pick by focus.
The social model reframes disability as a mismatch between a person and their environment, product, or society — not a deficiency in the person. As August de los Reyes put it:
"The biggest challenge is reframing disability as a mismatch between one's abilities and the environment. In other words, disability is designed."
The empowering corollary: if mismatches are designed, they can be un-designed. The same logic extends past disability to every exclusion in this skill — a form that rejects a valid name, an app that needs more bandwidth than a user can afford, copy only a fluent reader can parse. Each is a designed mismatch you can remove. Microsoft states it as: disability = mismatched human interactions.
A constraint is rarely permanent-only. Microsoft's Persona Spectrum maps each limitation across permanent, temporary, and situational states — which both grows the affected population enormously and reveals the universal need:
| Ability | Permanent | Temporary | Situational |
|---|---|---|---|
| Touch (one arm) | amputation | broken arm | a new parent holding a baby |
| See | blindness | cataracts / eye dilation | driving; bright sunlight |
| Hear | deafness | ear infection | a loud bar; a quiet library |
| Speak | non-verbal | laryngitis | a heavy accent on voice UI |
"We use the Persona Spectrum to understand related mismatches and motivations across a spectrum of permanent, temporary, and situational scenarios."
The scaling is the point: ~26,000 Americans a year experience permanent upper-limb loss, but counting temporary and situational impairments the number is more than 20 million. Designing for the edge serves the middle.
Go dimension by dimension and ask "who does this assume, and who does that leave out?" Each dimension has a reference file with the depth.
For the philosophy, principles, Persona Spectrum activities, and the accessibility/usability/inclusion distinction in full, see references/frameworks.md.
Inclusion is not a checklist you complete; it's a habit of noticing your assumptions. The fastest way to find exclusion is to stop designing for an "average user" (who doesn't exist) and instead pick a specific person at the edge — someone unlike you in language, income, device, or confidence — and walk your flow as them. The friction they hit is the design's, not theirs.
Primary sources: