When Can Accessibility Help? An Exploration of Accessibility Feature Recommendation on Mobile Devices
Jason Wu, Gabriel Reyes, Sam C. White, Xiaoyi Zhang, Jeffrey P. Bigham · 2021 · Proceedings of the 18th International Web for All Conference (W4A) · doi:10.1145/3430263.3452434
Summary
This paper tackles a significant gap in mobile accessibility: while smartphone operating systems now include dozens of built-in accessibility features — from screen readers and magnifiers to caption support and touch accommodations — most users are unaware these features exist or how to enable them. The researchers first conducted an online survey with 100 participants to quantify this awareness gap. Using a staged reveal methodology that avoided priming respondents, they found that only about 10% of responses correctly identified the relevant accessibility setting when presented with hypothetical scenarios involving vision, hearing, interaction, and mobility challenges. Only 18% of participants could correctly describe what "accessibility" meant in the context of their phone settings. The problem is especially acute among adults over 50, who are statistically more likely to experience age-related ability changes yet showed even lower awareness of available features. Building on these survey findings, the researchers developed a framework for "recommending accessibility" — proactively detecting when a user might benefit from an accessibility feature based on how they interact with their device. They categorized 48 iOS accessibility features and mapped them to four detection strategies: statistical detection (comparing usage patterns against baselines), near-miss detection (monitoring threshold failures like missed double-clicks), action sequence detection (identifying workaround behaviors), and grouped recommendations (suggesting related features based on features already enabled). They then built four prototype recommenders — for font size/zoom, subtitles and captions, side button click speed and AssistiveTouch, and grouped features — and tested them with two populations of older adults totaling 20 participants.
Key findings
The survey revealed a stark awareness gap: while 90% of participants in a baseline study of software engineers knew what accessibility features were, only 10.5% of older adults in the user study were aware of them — despite being the demographic most likely to benefit. Among all survey respondents, only 15.7% would have looked at their phone for a solution when facing an accessibility-related scenario, and just 12.1% identified the correct setting. In the user study with 20 older adults, the prototype recommenders triggered 19 recommendations total, of which participants rated 73.7% as useful, 21.1% as not useful, and 5.3% as neutral. The font size recommender was triggered for 3 participants (all wore glasses) who rated it 6.3/7 for usefulness. The click speed recommender was triggered most often due to near-miss detection; 70% of users who triggered it found it useful, rising to 100% among those who triggered the slowest threshold. Notably, 89.5% of participants were open to receiving accessibility recommendations, preferring low-frequency, non-intrusive methods like suggestions on the home screen or in messages. Several participants at a senior care residence had no concept of what "accessibility" meant in a computing context, even though they owned and used smartphones regularly.
Relevance
This research highlights a critical disconnect in digital accessibility: the features exist but users cannot find them. For practitioners, this underscores the importance of discoverability and proactive design — building accessibility features is only half the challenge; ensuring people know about and can activate them is equally vital. The framework of detection strategies (statistical, near-miss, sequence, grouped) provides a practical taxonomy that platform developers and app designers can use to think about surfacing accessibility options contextually. The finding that older adults who would most benefit from accessibility features are least aware of them has direct implications for device onboarding, setup wizards, and ongoing user support. The paper also raises important design considerations around how recommendations should be framed — the researchers deliberately avoided language like "it looks like you are having trouble" in favor of neutral phrasing like "did you know you can adjust the font size?" — a principle applicable to any accessibility-related notification or prompt.
Tags: mobile accessibility · accessibility features · older adults · recommender systems · feature discovery · aging · iOS accessibility · awareness
Standards referenced: iOS Accessibility API