Signature research case study · one complete HCD iteration
Designing a safer, evidence-led museum AR experience.
Wollaton Re:vived translates visitor research, environmental constraints and 30 traceable requirements into a handheld “Stop & Scan” experience. I led the research and requirements workstream within a collaborative team, then contributed to concept development, evaluation and the evidence-led roadmap.
Select the concept visual to inspect it at full resolution.
Design and evaluation contributor
One HCD iteration
BYOD mobile experience
Not launched or field-deployed
Section 1 of 12Overview
Overview
A promising concept — and an honest account of what remains unproven.
The project produced a research-informed AR concept, a formal requirements system and a mixed-method evaluation. It did not produce live visitor or business outcomes, so the results below are presented as formative prototype evidence rather than launch impact.
Functional and non-functional requirements with testable fit criteria.
Task-based, formative usability testing across four persona needs.
From first value and scanning to settings, quest suppression and facilities.
A promising prototype baseline, not evidence of live museum impact.
AR should return attention to the exhibit — not become another object demanding attention.
Research showed that visitors wanted visual, self-paced interpretation but resisted lengthy content, app-download friction, continuous phone holding and intrusive gamification. The strategy therefore moved away from constant immersion and toward short, stationary interactions with a clear exit, quick-view route and recovery path.
Visitor and environmental constraints
Interviews and Context of Use Analysis identified reading fatigue, glare, crowding, posture, connectivity and adoption barriers.
A bounded handheld interaction model
Three hero features, lightweight navigation, accessibility controls and explicit failure recovery formed one coherent concept.
Formative usability and expert review
Task data, think-aloud feedback, SUS, heuristic review and accessibility checks revealed both promise and material risks.
Context & challenge
The brief was not simply “add AR.”
Wollaton Hall operates as Nottingham’s Natural History Museum and welcomes almost 700,000 visitors annually. The client wanted a novel interactive layer that complemented its physical collection, supported a diverse audience and remained operationally feasible.
Static labels struggled to communicate movement, anatomy, scale and change over time.
The opportunity was not more information. It was a more legible relationship between the physical specimen and the story around it.
The digital layer could easily create a second source of fatigue.
Lengthy setup, raised-arm use, walking while viewing and unreliable tracking could make AR less inclusive than the labels it aimed to improve.
How might we reveal meaningful information about physical exhibits through short, self-paced interactions that remain safe, inclusive and operationally self-service?
Constraints became product inputs
Heritage environment
The digital concept could not require physical alterations to the Grade I listed building.
High concurrent use
The concept needed to scale without headset inventory, fitting, cleaning or specialist supervision.
Diverse visitors
Age, language, vision, hearing, mobility, confidence and visit context varied substantially.
Crowded physical space
Interactions had to be usable while stationary and resilient to interruption.
Unreliable connectivity
Core stop content needed a low-bandwidth or offline-capable delivery model.
Attention is finite
The experience needed to minimise screen time and return attention to the physical collection.
Role & ownership
I led the research and requirements workstream — not every part of the project.
Clear attribution matters in collaborative case studies. My strongest ownership was establishing a consistent research process and translating evidence into requirements. Prototype production and final visual design were team-owned, with another teammate spearheading the interface build.
Research operations and requirements
- Interview protocol, script and field guide
- Interviewer and scribe responsibilities
- Primary-research execution
- User requirements gathering and specification
- Evidence-to-requirement traceability
Product direction and evaluation
- Empathy maps and persona synthesis
- Concept ideation and design critique
- Evaluation plan and risk review
- User and non-user evaluation
- Findings synthesis and recommendations
Prototype and final submission
- Overall product direction
- Low- and high-fidelity design production
- Interactive prototype assembly
- Wider report development
- Final group reflection and submission
Standardising distributed interviews without pretending every session was identical.
Before: neutral wording, consent, roles and scenario order.
During: one moderator, one scribe, behaviour and sentiment capture.
After: a five-minute hot wash and shared evidence format while memory was fresh.
Senior-level storytelling is not achieved by claiming the whole product. It comes from showing where the workstream changed the quality of team decisions and where ownership was genuinely shared.
Goals & measures
Success was defined before the prototype was judged.
The evaluation strategy was defined around adoption, comprehension, comfort, safety and recovery — not visual preference alone.
Reach first value quickly
Minimal onboarding, no required account and a lightweight route to the first supported stop.
Make one key idea memorable
Plain language, visual interpretation and progressive disclosure should reduce reading fatigue.
Support quick and deep use
Visitors should control pace, modality, quest participation and when to leave AR.
Keep core interactions stationary
Stop & Scan zones, visible exit controls and lightweight wayfinding reduce walking-while-viewing risk.
Recover without staff support
Tracking failures should produce immediate guidance and an alternate demo route.
Preserve readability and modality choice
Text size, contrast, dark mode, reduced motion, captions and audio controls should remain independent.
Discover
Research examined both the visitor and the museum environment.
The product would be used while standing, walking, carrying belongings, managing children and navigating crowded galleries. Designing only for the phone screen would have missed the most consequential constraints.
Semi-structured interviews
Short sessions explored museum habits, content fatigue, personal-device AR, comfort, accessibility and reactions to the proposed feature concepts.
Context of Use Analysis
The team mapped lighting, posture, crowding, interruption, connectivity, staff support and safety as product requirements rather than edge cases.
Literature and precedent review
Progressive disclosure, cultural-heritage AR, human-centred design and accessibility guidance challenged unsupported assumptions.
Four themes changed the direction of the product
Information overload and reading fatigue
Visitors skimmed headings and abandoned dense labels. The product response was visual-first interpretation with a concise default and optional depth.
Technical friction and physical ergonomics
Download size, poor signal, battery use and continuous raised-arm interaction threatened adoption. The concept moved toward short bursts, offline-ready content and Glance Mode.
Pacing and visitor control
Participants valued controlling the Time-Slice sequence rather than watching a fixed animation. The product made skimming and deeper exploration equally valid.
Gamification was polarising
Families saw value while solo adults anticipated annoyance. The quest became bounded, clearly optional and suppressible.
Persona system: four competing definitions of a “good visit”
The persona boards came from the supplied coursework PDFs. Each is presented as a research synthesis artifact, not as a claim of statistically representative segmentation.
Readable, independent interpretation
Needs adjustable text and contrast, visual storytelling and alternative formats without repeatedly asking staff for help.
Engagement without more cognitive work
Needs child-friendly storytelling, safety, clear facilities and an activity that does not make the adult manage another complex system.
Confidence beyond language-heavy labels
Needs plain language, visual interpretation, localisation and a memorable route through unfamiliar content.
Self-paced depth without social friction
Needs clear content, control over pace and a discreet interaction that can be exited or ignored at any time.
Empathy maps retained the behavioural evidence behind the profiles
Context of use was the decisive design input
| Context factor | Observed UX risk | Product implication |
|---|---|---|
| Lighting and glare | Physical signs and mobile overlays can become unreadable. | High-contrast content, dark mode and adjustable typography. |
| Standing posture | Continuous raised-arm use creates physical fatigue. | Short stop interactions, Glance Mode and post-scan review. |
| Crowding and movement | Walking while viewing creates collision and attention risk. | Stationary Stop & Scan zones and no real-time AR arrows. |
| Connectivity | Poor signal can block setup and content loading. | Offline-ready core stops and a constrained package size. |
| Interruption | Visitors may need to pause for queues, children or navigation. | Persistent state, clear exit and recoverable short interactions. |
| Staff capacity | A high-support concept cannot scale to peak visitor volume. | Self-service setup, visible recovery and demo fallbacks. |
Participants were friends, family and university peers rather than the visiting public. The persona and context work improved decision quality, but future field research must include real families, older visitors and people with diverse access needs.
Define
Evidence became principles, requirements and explicit exclusions.
Create a handheld interpretation layer that reveals information unavailable from static labels while preserving visitor control, physical safety and museum operational feasibility.
Five principles governed design trade-offs
Reveal, do not duplicate
AR should expose anatomy, movement, scale or historical change that the physical object alone cannot show.
Progressive disclosure by default
Every stop begins with one concise idea, then allows the visitor to choose deeper content.
Stationary, short and interruptible
Core tasks can be completed without walking and without holding the phone up continuously.
Optional features stay optional
Gamification, audio and advanced detail never block the primary learning route.
Failure is part of the journey
Tracking loss, poor signal and uncertainty receive visible guidance and a self-service fallback.
The “golden thread” connected research to evaluation
| Evidence | Requirement response | Designed evidence | Evaluation signal |
|---|---|---|---|
| Visitors skip long text. | Plain-language summaries and layered detail. | Glance Mode and progressive Reveal cards. | Key-idea recall, clarity and deep-dive access. |
| Phone holding causes fatigue. | Short-burst interactions and persistent exit. | Stop & Scan, quick-view route and bounded sequence. | Task time, comfort feedback and exit speed. |
| Tracking failure can cause abandonment. | Feedback within three seconds and demo fallback. | Scanning status, recovery guidance and alternate route. | Error count, recovery success and frustration. |
| Quest appeal differs by visitor type. | Bounded opt-in quest with persistent suppression. | Family Quest entry, progress and reminder controls. | Opt-in/out success and annoyance rating. |
| Lighting and language reduce comprehension. | Contrast, text controls, plain language and localisation model. | Settings, visual-first content and language-ready structure. | Accessibility review and proxy comprehension. |
Explicit exclusions protected the iteration
Head-worn hardware
Inventory, fitting, charging, cleaning and throughput conflicted with operational scale.
Real-time indoor AR navigation
Technical dependence and walking-while-viewing risk outweighed its visual appeal.
Mandatory account creation
One-time visitors should reach value without a privacy or setup barrier.
Mandatory gamification
Adult and solo visitors could ignore the quest without repeated prompts.
Develop
Product strategy was a series of trade-offs, not a straight line to polished screens.
Choose handheld BYOD over head-worn AR.
Familiar personal devices improved theoretical throughput and reduced museum responsibility for hardware, hygiene and fitting — while introducing download, battery and arm-fatigue risks that had to be designed around.
Remove real-time AR wayfinding from scope.
The team replaced a compelling but risky concept with stop pre-selection, facility support and a conventional map.
Three hero features concentrated the prototype
Scan & Reveal
Reveal anatomical and interpretive information that adds meaning to a physical specimen.
Time-Slice
Move through discrete historical stages at a visitor-controlled pace.
Family Quest
Offer a bounded three-stop activity that never blocks the core experience.
Ideation focused on architecture before visual polish
Bound the screen-heavy interaction to a clear stop-based loop.
- Start with value and choice.
- Explain safety and camera use.
- Select a nearby supported stop.
- Scan while stationary.
- Explore at a chosen depth.
- Exit and return to the physical visit.
Fast start, safe onboarding and transparent permissions
Onboarding was limited to three screens: value, stationary safety and camera/privacy. Returning visitors could bypass it.
Pre-select stops instead of asking visitors to navigate through the camera
The supported-stop model reduced wandering and made lightweight maps and events accessible independently of AR.
Use progressive disclosure to preserve depth without recreating the label wall
Scanning leads to an interpretive overlay, then Glance or Time-Slice routes let visitors control attention, pace and depth.
Keep the quest bounded and accessibility controls independent
Families could opt into the activity, suppress reminders and reach a clear ending. Display and modality settings remained useful without the quest.
Design tracking recovery as a first-class flow
The system describes what went wrong, offers corrective steps and preserves a demo fallback rather than leaving the visitor at a dead end.
Structure the recovery logic
Non-blaming status, concrete tips and a route out.
Make the next action unmistakable
High-contrast guidance and a visible demo fallback reduce abandonment risk.
Evaluate
The evaluation was designed to find risk, not confirm that the interface looked finished.
The evaluation triangulated task performance, think-aloud evidence, SUS, heuristic inspection, cognitive walkthrough and an accessibility checklist. The objective was to surface risk before field deployment.
University peers acted as proxies for solo, family, international and accessibility-first needs.
The protocol covered first value, content, scanning, pacing, settings, safety, quest control and facilities.
Heuristic review, cognitive walkthrough and accessibility checks complemented the user sessions.
Evaluation task matrix
| Task | Scenario | Success signal | Primary evidence |
|---|---|---|---|
| T1 | Reach first meaningful content. | Complete without assistance. | Time, errors and help requests. |
| T2 | Use Glance Mode in a crowded-gallery scenario. | Find the route and state one key idea. | Time, recall and perceived usefulness. |
| T3 | Trigger Scan & Reveal. | Reach the overlay and identify a new insight. | Success, errors and perceived meaning. |
| T4 | Skim and deeply explore the Time-Slice. | Move between both pacing modes. | Time, comprehension and satisfaction. |
| T5 | Change display and modality preferences. | Toggle at least two settings. | Time, success and perceived access. |
| T6 | Exit AR and return to the museum view. | Leave in one action without intrusive cues. | Exit time, control and intrusiveness. |
| T7 | Choose or suppress the Family Quest. | Ignore or disable it without repeated prompts. | Success, persistence and annoyance. |
| T8 | Find a facility and next stop. | Use the map or help card successfully. | Time, success and usefulness. |
Risk management changed the prototype before testing
Adoption friction
Offline-ready core content, minimal onboarding and no login addressed Wi-Fi, storage and setup resistance.
Crowding safety
Stationary Stop & Scan use, subtle cues and a persistent exit reduced walking-while-viewing risk.
Accessibility gaps
Captions, text controls, contrast, dark mode and reduced motion became explicit evaluation criteria.
Tracking failure
Immediate feedback, recovery steps and a demo mode protected the core experience.
Phone fatigue
Short interactions and quick-view content reduced continuous camera use.
Information overload
Plain language and layered content prevented AR from becoming another wall of text.
The participant pool was small, young and digitally literate. Task and expert data can guide another iteration, but they cannot establish accessibility, adoption or safety in real museum conditions.
Results
Completion looked excellent. The friction underneath it was more useful.
All eight participants completed all eight prototype tasks. That headline alone would have hidden the errors, hesitation, accessibility failures and adoption concerns that shaped the next iteration.
A promising formative baseline from eight proxy participants.
Completion did not mean the interactions were friction-free.
Performance data revealed where hesitation concentrated
| Task | Average time | Errors | Satisfaction | Interpretation |
|---|---|---|---|---|
| T1 · First value | 7.8s | 0 | 4.6 / 5 | Fast start was broadly understood. |
| T2 · Glance Mode | 6.2s | 0 | 4.5 / 5 | The route was quick, but terminology still caused questions. |
| T3 · Scan & Reveal | 10.4s | 1 | 4.6 / 5 | The camera-style control created ambiguity. |
| T4 · Time-Slice | 28.2s | 0 | 4.2 / 5 | The richest task needed clearer progress guidance. |
| T5 · Preferences | 8.4s | 0 | 4.6 / 5 | Settings were discoverable in the prototype. |
| T6 · Exit AR | 5.1s | 0 | 4.8 / 5 | Visible control supported confidence and quick disengagement. |
| T7 · Quest control | 11.2s | 3 | 4.0 / 5 | The optional feature produced the most errors and lowest satisfaction. |
| T8 · Facilities | 6.8s | 0 | 4.7 / 5 | Lightweight navigation was easier to understand than AR wayfinding. |
The qualitative friction was more valuable than the completion rate
“Glance Mode” described a mechanism, not a benefit.
Participants could complete the task but still questioned what the mode meant and why audio was absent.
The scan control implied a manual camera action.
A familiar shutter icon contradicted an automatic detection model, creating a severity-three match-to-real-world issue.
The map palette failed the accessibility review.
Thematic colours did not meet the intended contrast threshold, becoming a severity-four issue before any public deployment.
Quest discoverability and suppression needed clearer guidance.
The optional feature generated the highest error count and required a one-time explanation without recurring prompts.
Prototype usability did not solve download reluctance.
Participants questioned installing a native application for a single visit, exposing a product-distribution problem outside the clickable flow.
International visitors needed real localisation.
Plain English helped, but proxy feedback showed that a language toggle — beginning with Mandarin — remained a high-value next step.
Next iteration
Evaluation findings became a prioritised design and product roadmap.
These changes are not presented as validated outcomes. They are proposed responses that require another design-and-test cycle.
Clarify automatic scanning
Remove the shutter metaphor, use an active reticle and add haptic or audio confirmation.
Rebuild map contrast and hierarchy
Prioritise legibility under representative lighting before preserving thematic colour.
Rename and explain quick-view behaviour
Use benefit-led language and a one-time tooltip for the optional quest.
Implement localisation and multimodal feedback
Establish a scalable language model and reduce reliance on visual-only scanning status.
Persist content after a successful scan
Let visitors lower the device and review interpretation without maintaining camera alignment.
Investigate frictionless delivery
Compare native installation with WebAR, App Clips and Instant Apps in an adoption experiment.
Direct response to the scanning-affordance finding
Camera control suggested “tap to capture.”
The visual language conflicted with automatic marker detection.
Reticle and status communicate active scanning.
This direction still requires retesting with multimodal confirmation and tracking recovery.
Outcomes & limits
The outcome was reduced uncertainty — not a claim of transformed visitor behaviour.
The core interaction model was usable enough to evaluate seriously.
The task flow and SUS baseline supported another iteration rather than a reset.
Accessibility, adoption and ergonomic risks were visible before deployment.
Contrast, download reluctance, quest confusion and tracking affordance emerged before engineering investment.
Research removed technically impressive but misaligned scope.
The team excluded headsets, mandatory accounts, constant immersion and real-time AR navigation.
The project did not validate the museum.
- Real tracking, glare, noise, connectivity and crowd behaviour
- Families with young children and visitors with substantial access needs
- Download conversion, permissions and abandonment
- Staff support burden, dwell time, return visits or commercial outcomes
Instrument the entire value chain.
- QR-to-launch conversion and time to first content
- Scan success, recovery and key-idea recall
- Comfort, access-setting use and safety incidents
- Quest participation and suppression behaviour
- Staff intervention and content-maintenance effort
Wollaton Re:vived is an evaluated design proposal, not a launched museum product. Its strongest outcome is a defensible chain from visitor evidence to product boundaries, testable requirements and a prioritised next iteration.
Reflection
The most valuable design work happened before and after the polished screens.
Requirements stopped being documentation created after research. They became the mechanism for resolving disagreement, controlling scope and judging whether a design decision was justified.
Personal learning from the project
A method is valuable when other people can use it consistently.
The field guide, shared roles and hot-wash process mattered because interviews were distributed across a team.
Ambition must survive the context of use.
Removing AR wayfinding produced a less theatrical but safer, clearer and more testable product direction.
High completion can coexist with meaningful friction.
The most useful evidence came from hesitation, terminology questions, contrast failures and adoption concerns.
Leadership does not require overstating authorship.
My ownership was strongest in research and requirements; prototype production and visual execution remained collaborative.
Bring field validity and accessibility testing forward.
A future iteration should test in the museum with representative families, older visitors and people with diverse access needs.
Make the evidence architecture visible.
Decision logs, traceability and explicit claims boundaries make a complex research case study easier to trust and maintain.
This case study is based on the Group D university report and supplied project artifacts. It separates observed evidence, collaborative outputs and proposed future work, and does not claim that the experience was deployed at Wollaton Hall.
Need a guided walkthrough of the work and decision-making?
Start a conversation