An Accessibility Usability Study and Redesign of the UMANG App for Low-Vision and Older-Adult Users

Access Denied? Redesigning UMANG for Low-Vision and Older-Adult Users is a B.Des UX graduation project that studies a real, high-stakes artefact: UMANG (Unified Mobile Application for New-age Governance), the Government of India's single app for EPFO, pension, utility-bill and hundreds of other citizen services, built by MeitY's National e-Governance Division. The people who most need these services — older adults and users with visual disabilities — are the least studied by its interface.

The project runs in two linked tracks. The audit track tests UMANG's Android app against the WCAG 2.1 AA success criteria and the Government of India's GIGW 3.0 guidelines (which align with IS 17802), including a full TalkBack screen-reader walkthrough of the login, service-discovery and EPFO passbook flows. The study track runs task-based moderated usability tests with 14 ethically recruited participants — 8 older adults and 6 low-vision screen-reader users contacted through a district blind-welfare association and a senior-citizens' forum — capturing task success, time, SUS scores and severity-rated findings.

The synthesis drives a redesign of three flows in Figma (onboarding, service discovery, EPFO passbook viewing) with contrast fixes, dynamic type, proper element labels, focus order and plain bilingual copy. A five-participant validation round measures whether the fixes moved the numbers. The deliverables — process document, audit matrix, prototype and portfolio case study — go to a jury that evaluates research, process, iterations, final artefact and presentation.

Government services are moving to apps faster than citizens are moving to apps. UMANG concentrates EPFO, pension and bill services in one place, but the citizens with the greatest need — older adults and people with visual disabilities — face unlabelled controls that defeat screen readers, small touch targets, low-contrast cards, English-heavy jargon and an onboarding flow built around OTP juggling. None of this is visible in a compliance document; it only shows up when a real user tries to view their own passbook and fails. There is no published, evidence-based account of how these two user groups actually perform on UMANG, and no demonstrated redesign that shows what conformance and plain design would look like in practice. A rigorous, ethically conducted usability study of the real app, followed by a tested redesign, gives both the design community and the standards conversation the missing evidence.

Abstract

This graduation project asks a narrow, testable question: can a first-time low-vision user or a 65-year-old actually complete core tasks on UMANG, India's flagship government super-app? It answers with evidence — a WCAG 2.1 AA and GIGW 3.0 audit, a TalkBack walkthrough, and a 14-participant task-based usability study — and then redesigns the three weakest flows in Figma and retests them.

Introduction

UMANG aggregates 2,500+ services from the centre, states and local bodies: EPFO passbook and claim status, pension services, bill payments, DigiLocker access. Its users include millions who do not read English comfortably, many whose eyesight fails before their phone skills do, and screen-reader users for whom an unlabelled button is a dead end. Accessibility for Indian government apps is not optional: GIGW 3.0 makes it a mandated guideline set, aligned with IS 17802. Yet the lived experience of the least fluent users is rarely documented in student research.

Existing situation and the gap

• UMANG's public documentation positions it as citizen-first, but no published, evidence-based usability account exists for low-vision and older-adult users.
• Prior student and industry studies in this space either audit compliance on paper or test young, confident users.
• The gap this project fills: combining a standards audit with observed task performance of the actual user groups, then showing a redesign that closes the measured gaps.

The proposed work

1. Audit: automated checks plus a manual WCAG 2.1 AA matrix and a GIGW 3.0 mapping over the Android app; a complete TalkBack walkthrough of three flows.
2. Study: 14 participants (8 older adults 60–70, 6 low-vision screen-reader users), 5 tasks each (install and register, find a service, view EPFO passbook, check claim status, raise and track a query), moderated think-aloud sessions with screen recording.
3. Synthesis: severity-rated findings (Nielsen's 0–4 scale), SUS scores per group, affinity mapping into a problem tree.
4. Redesign: Figma rebuild of onboarding, service discovery and passbook viewing — contrast, dynamic type, element labels, focus order, touch-target size, plain bilingual copy.
5. Validation: retest with 5 participants against the same task set and SUS.

Feasibility

UMANG is free and runs on any mid-range Android phone; TalkBack is built into Android; the standards are public documents. Recruitment runs through two partner organisations (a district blind-welfare association and a senior-citizens' forum), whose coordinators co-sign the ethics plan. Sessions need a quiet room, a recorder and about 60 minutes each. The main risks — recruitment slippage and consent friction — are absorbed by a two-week buffer in the schedule and oral consent in the participant's language, recorded and archived with written forms.

In scope

• The Android app of UMANG (current public version) on two mid-range devices of different screen sizes; the web portal is referenced only where a flow differs materially.
• Three flows end to end: registration/onboarding, service discovery, EPFO passbook viewing; other services are touched only during free exploration.
• Two user groups: older adults (60–70) who use a smartphone at least weekly, and low-vision screen-reader users; both recruited through partner organisations with written consent.
• A WCAG 2.1 AA + GIGW 3.0 audit matrix, a usability study with SUS and severity ratings, a Figma redesign of the three flows, and a 5-participant validation round.

Out of scope

• iOS parity, the full 2,500-service catalogue, and back-end or server behaviour.
• Producing or shipping code to UMANG/NeGD — this is an independent academic redesign, not an official collaboration.
• Cognitive-disability and motor-accessibility user groups (documented as future work).
• Any collection of participants' government ID numbers, EPFO credentials or actual account balances.

The project follows the Double Diamond as its research-through-design frame, with the first diamond running the evidence tracks and the second running the redesign.

Sampling: purposive recruitment through a district blind-welfare association and a senior-citizens' forum, with snowball extension to reach 14; the pilot session with 1 older adult tests the task scripts. Instruments: consent forms in the participant's language, task scripts, observation sheets, the SUS questionnaire, the Nielsen 0–4 severity scale and the audit matrix template. Analysis plan: SUS scored per the standard method (benchmark 68); task success and time compared across groups; findings severity-rated and mapped to WCAG criteria; affinity clusters reduced to a problem tree that drives the redesign backlog. Ethics: written + recorded oral consent; participation is voluntary with a token of appreciation; no government ID numbers, credentials or balances are collected or recorded; participants are identified only as P01–P14 in all documents; recordings stay on an encrypted personal drive and are deleted after jury submission; the protocol is approved by the programme's academic review and co-signed by the partner coordinators.

The project is a study system rather than a software build, so the architecture below shows the study design — how evidence flows from the two tracks into the synthesis that drives the redesign, and back into validation.

flowchart TD
  A["Recruitment through partner<br/>organisations + ethics paperwork"] --> B["Pilot audit and pilot session"]
  B --> C1["Track 1: standards audit<br/>WCAG 2.1 AA + GIGW 3.0 matrix"]
  B --> C2["Track 2: usability study<br/>14 participants, 5 tasks each"]
  C1 --> D1["TalkBack walkthrough log<br/>of 3 flows"]
  C2 --> D2["Recordings, task success,<br/>time-on-task, SUS"]
  D1 --> E["Severity-rated findings<br/>Nielsen 0-4 scale"]
  D2 --> E
  E --> F["Affinity mapping: personas,<br/>journey maps, problem tree"]
  F --> G["Redesign in Figma:<br/>onboarding, discovery, passbook"]
  G --> H["Validation retest:<br/>5 participants, same metrics"]
  H --> I["Before/after report,<br/>process document, jury deck"]

Each usability session itself follows a fixed protocol, so that data is comparable across all 14 participants:

flowchart LR
  A["Consent +<br/>context questions"] --> B["Training task:<br/>find weather service"]
  B --> C["5 scored tasks,<br/>think-aloud"]
  C --> D["SUS +<br/>post-task ratings"]
  D --> E["Debrief +<br/>token of appreciation"]

The mapping is deliberate: every severity-rated finding carries its WCAG criterion, its audit-matrix row and the redesign decision it produced, so the jury can trace any interface change from raw evidence to final artefact in one step.

• Extend to cognitive-accessibility and motor-accessibility user groups with adapted protocols and partners.
• Localisation testing in two more Indian languages with community partners.
• A working HTML/ARIA reference build of the redesigned passbook flow to demonstrate the behaviours concretely.
• A comparative study of two more government super-apps using the same audit matrix, building a state-of-practice report for Indian public-interest UX.