Accessibility in Game Design
Practical game accessibility: remapping that works, captions, contrast, difficulty options, photosensitivity, and the guidelines worth building against.
By Velloreth
Accessibility in games has a reputation as a compliance exercise. It is not. It is configuration design: deciding which parts of your game’s behaviour a player is allowed to change, and then making sure those settings actually work.
The features that matter most are ones the majority of players use anyway. Field of view sliders, subtitle options, remapping, audio mix controls, difficulty settings — these are ordinary settings screens. The accessibility question is whether a player who must change them can, not whether anyone will.
Microsoft’s Xbox Accessibility Guidelines (XAGs) make this framing explicit: they are a set of 23 best practices developed with the Gaming & Disability Community, intended as design prompts and test checklists, and deliberately not a compliance or legal standard. That is the right posture — a catalogue of known barriers with known solutions, not a box to tick.
The barrier map
The XAGs are organised by barrier type, and the grouping is a useful way to audit your own game:
| Area | Guidelines | The practical question |
|---|---|---|
| Text and contrast | 101, 102 | Can it be read at a distance, on a TV, at default size? |
| Visual and audio channels | 103, 104, 105, 106, 111 | Does every cue exist in more than one channel? |
| Input | 107 | Can the game be played with the controls this player has? |
| Difficulty and pacing | 108, 116 | Can the challenge and the clock be changed? |
| Clarity and distraction | 109, 117, 118 | Is the important thing visible, and is the flashing safe? |
| Interface | 112, 113, 114, 115 | Can menus be navigated and understood without sight of a cursor? |
| Communication | 119, 120 | Can everyone in the match talk, in a form they can use? |
| After sales | 121, 122 | Can players find the features, and get help? |
Nothing on that list is exotic. Each row describes a failure mode that shows up in shipped games constantly.
Input: remapping is the beginning, not the end
XAG 107 is the one most teams believe they have already solved, and the one that most often fails in practice. Full remapping of every action to every button is necessary and insufficient. What decides whether a remap helps:
Hold versus toggle. A player who cannot hold a trigger needs every “hold to aim” action available as a toggle. Same for sprint, crouch, and stick-click actions — L3 and R3 are inaccessible to many players precisely because they require sustained pressure on a stick that must also be moved.
Simultaneous input requirements. Anything needing two buttons at once (sprint-plus-crouch, modifier keys on PC) is a hard barrier. Provide alternate single-input routes for every critical action.
Analog dependence. If a menu can only be navigated with a mouse or a thumbstick, a player using single-press digital buttons cannot reach the settings screen at all — which means they cannot reach the settings that would have helped them. Menus need digital navigation paths; Grounded’s arrow-key look controls are the canonical example of solving this properly.
Sensitivity and deadzone, per control. Not one global slider — separate horizontal and vertical sensitivity, stick deadzones, and aim acceleration, because the physical constraints that make one necessary rarely apply uniformly.
Prompt design. Button prompts must reflect the current mapping, not the default. A remapped control scheme with default glyphs is a worse state than no remapping at all.
Text, contrast, and visual clarity
TV-distance viewing changes the thresholds. Text that passes on a monitor at 60 cm fails on a couch at 3 m.
- Minimum default sizes, with player scaling beyond them (XAG 101). Subtitles, HUD labels, and objective text all count — a scaling option that covers only dialogue leaves half the interface unreadable.
- Contrast. The web standard is a useful anchor: 4.5:1 for normal text, 3:1 for large text and for non-text UI elements and meaningful graphics. Over gameplay scenes, verify against the brightest background the text can appear over, not a static mockup.
- Text background plates. A subtle scrim behind subtitles raises contrast without redesigning the palette, and is the cheapest subtitle improvement available.
- Colour is never the only channel (a principle shared with WCAG SC 1.4.1). Team colours, health states, interactable highlights, and puzzle solutions all need a second signal: shape, icon, pattern, label, or position. A dedicated “colourblind mode” that recolours the palette helps some players and does nothing for state distinctions carried by colour alone.
- Object clarity (XAG 109). Enemy outlines, interactable highlights, and a configurable reticle are not assists bolted onto a pure experience; they are how you make the game’s information legible on a small screen in a bright room.
Audio: every cue needs a second channel
Sound carries enormous gameplay information — footsteps, reloads, enemy callouts, objectives. A player who cannot hear it is playing a different, harder game.
Subtitles and captions (XAG 104) should be on by default, with: speaker identification, size and background options, and — the part almost always missed — captions that describe significant non-speech sounds. [footsteps above], [alarm], [gunfire, distant]. Closed captions and subtitles are not the same feature; one is dialogue, the other is the sound mix rendered as text.
Per-channel volume (XAG 105). Music, dialogue, effects, UI, and voice chat as separate sliders. A player who needs dialogue louder to follow the story, or effects quieter because of sensory overload, cannot do either with a single master volume.
Visual alternatives for audio cues. Visual sound indicators, directional subtitles, and optional haptics (XAG 110) for impacts and low-health states. Sea of Thieves’ chat wheel is the same idea from the communication side: a text-based route into a voice-shaped interaction.
Screen narration (XAG 106) — full UI text-to-speech — is a larger build. It is also the difference between a game being configurable and being navigable for blind and low-vision players, and platform support now makes it less expensive than it once was.
Difficulty, time, and cognitive load
Difficulty options (XAG 108) exist so that a player blocked at one encounter can continue to the rest of the game. Minecraft’s four Survival presets plus creative mode is the reference implementation of the principle that there should be enough distinct settings that different barriers are covered.
What makes difficulty settings work:
- Changeable at any time, without progress loss. Locking difficulty at the start, or punishing a change, means the setting cannot be used by the player who discovers mid-game that they are stuck.
- Granular, not just a single slider. Separate damage taken, damage dealt, resource availability, and puzzle timing. A player with a motor constraint does not need easier puzzles; a player with a cognitive load issue does not need slower combat.
- Not just numbers. Difficulty can change assist behaviour — aim assist strength, quick-time-event skipping, auto-triggers — without touching combat balance at all.
- Time limits (XAG 116) get the same treatment: adjustable, pausable, or removable, because a clock adds a barrier that has nothing to do with whether the player understood the task.
Motion and photosensitivity (XAG 117, 118) round this out. A reduced-motion option that disables screen shake, chromatic aberration, motion blur, and camera bob is now table stakes; many players without any diagnosed condition turn these off anyway. Flashing content should stay well below the well-known three-flashes-per-second threshold, and the most aggressive effects — full-screen strobes, hard camera snaps — should be optional.
The cheap, high-value list
Ordered roughly by impact per unit of effort for a small team:
| Feature | Why it matters | Cost |
|---|---|---|
| Full remapping, all actions | Unlocks the game for motor-impaired players | Low — settings work |
| Hold-to-toggle options | Same, for sustained inputs | Low |
| Subtitles on by default, size options | The most-used accessibility setting in games | Low |
| Speaker IDs and sound captions | Makes the sound mix legible | Low–medium |
| Per-channel audio sliders | Solves several barriers with one screen | Low |
| Scalable HUD and text | TV and low-vision viewing | Medium |
| Reduced motion / shake toggle | Broad appeal beyond disability | Low |
| Colour plus shape/icon for states | Fixes colour-blind play properly | Medium |
| Difficulty changeable mid-game | Prevents hard blocks | Low |
| Consistent UI navigation order | Screen reader and motor players both | Medium |
None of these require new gameplay systems. Almost all of them are settings screens, default values, and one-time design rules about how states are rendered.
Finding the barriers you cannot see
Checklists catch the known failures. They do not catch the ones specific to your game — the puzzle that requires rapid colour matching, the tutorial that only demonstrates a mechanic visually, the boss whose tells are a subtle audio cue.
Two ways to close that gap:
Structured internal review. Walk the XAG list against a real build once per milestone, not once at the end. The later you look, the more of the list becomes “would require a redesign” rather than “requires a settings option.”
Play with the people who need the features. Community programs — Microsoft’s Xbox Accessibility Insiders League is one, and disability gaming communities run playtests of their own — recruit players who will tell you exactly where the game broke for them. Developers who have done this consistently report that the feedback changes design decisions, not just flags bugs: an encounter redesigned because the assumed cue could not be perceived is a better encounter for everyone.
Documentation is a feature
XAG 121 is the guideline teams skip, and it undoes work that was already done. Accessibility features that players cannot find are features that do not exist.
- List them on the store page, specifically. “Remappable controls” and “Subtitle size options” are searchable facts, not marketing.
- Use platform accessibility metadata and store tags so players can filter for what they need.
- Surface the settings at first launch. Minecraft Dungeons opening directly onto an accessibility configuration screen — and reading it aloud automatically when the platform’s screen-reader setting is on — is the pattern: configure once, play forever.
- Keep an accurate public list. Overclaiming in accessibility metadata is worse than saying less, because a player plans their purchase around what you wrote.
The frame worth keeping
Accessibility work fails when it is framed as diluting a vision. It succeeds when it is framed as configurability: which parts of this game’s behaviour should a player be able to change, and what does changing them require from us.
Start with the settings screen you already have. Every control in it is a decision about who can play. Make those decisions on purpose, document the ones you make, and let the guidelines — the XAGs, the Game Accessibility Guidelines, or both — do what they were built to do: give you the list of barriers other players have already paid to discover, so you do not have to.