Turning Overwhelming Camera Rolls
into Libraries of Meaningful Memories
A capstone UX research and design project exploring how to reduce decision fatigue in photo curation — and help people build photo libraries centred on what truly matters to them.
Project Overview
The brief
My Role
User Researcher
Timeline
10 Weeks
Methods
Literature Review, Competitive Analysis, Semi-Structured Interviews with Contextual Inquiry, Affinity Diagramming, Multifidelity Prototyping, RITE Method, Usability Testing
Tools
Figma, FigJam, Zoom
Focus Area
Photo Curation, Memory & Meaning, Decision Fatigue, AI-Assisted UX
Most people have thousands of photos on their phones and almost no system for deciding which ones actually matter. Existing apps like Google Photos and Apple Photos nudge users toward buying more storage rather than curating more thoughtfully. Storage-focused tools like SwipeWipe identify near-duplicates but none of them help users ask the harder question: does this photo hold long-term sentimental value?
Memento was born from a shared brainstorming session where photo curation was unanimously identified as something overwhelming and tedious. Through a full research-and-design cycle — from literature review through high-fidelity usability testing — the team designed a photo-curation system that treats every kept photo as a memory, not just a file. The word "content-aware" became the early differentiator, with AI helping determine what users are most likely to want to keep.
Project Goals
What we set out to achieve
The project was guided by two primary goals, each with defined measures and documented revisions as the project evolved.
Reduce Decision Fatigue During Photo Curation
Reduce the perceived effort of photo curation through prompting, AI-powered suggestions, and gentle nudging — making the process feel manageable rather than overwhelming.
Revised to: Observational data and explicit open-ended questions. A direct "decrease" comparison would require a control condition we didn't have the resources or sample size to make meaningful.
Empower Users to Build a Gallery of Meaningful Photos
Help users develop a photo library where every kept image is a memory — not just an unreviewed file — by giving structure to the abstract process of identifying what matters.
Revised to: Tested earlier in design interviews by asking users if they were satisfied with the photos they chose. Personal photo use requires IRB approval, so stock photos and team member photos were used instead.
Methodology
How we approached this
Literature Review
Surveyed research on decision fatigue, photo curation processes, and the psychology of digital decluttering. Used the DePaul Library to access publications. Key finding: voting is perceived as less burdensome than deleting — directly shaping our interaction model. Also found research on why swipe-based models don't always work for curation tasks.
Competitive Analysis
Examined Apple Photos, Google Photos, SwipeWipe, Clean My Phone, and Remo. Reviewed app stores and published articles comparing photo curation tools. Documented each app's features, interaction model, and marketing focus to identify where Memento could differentiate.
Semi-Structured Interviews with Contextual Inquiry
Conducted 7 interviews (ages 25–32) over Zoom via word-of-mouth recruitment. One team member moderated; the other took notes. Covered photo-taking habits, curation triggers, organisational methods, and emotional relationships with photos. The contextual inquiry component — asking users to find 10–20 photos from their own camera roll and discuss them live — was added mid-plan for richer, more grounded data.
Affinity Diagramming via FigJam
Transferred notes from all interviews into FigJam and clustered them into themes. Atlas.ti was removed from the original plan due to the significant time commitment of full coding. Notes-based diagramming was sufficient for the project's scope and allowed faster movement into prototyping while insights were still fresh.
Multifidelity Prototyping — Sketching & Figma
Began with pen-and-paper sketches to explore multiple layout and interaction patterns without committing to visual detail. After identifying common patterns, recreated these in low-fidelity grayscale Figma screens to validate structure and flow. Once tested, produced the final high-fidelity prototype with consistent icons, styling, and components for Round 2 of testing.
Design Interviews — RITE Method
Used the Rapid Iterative Testing and Evaluation method with low-fidelity paper prototypes in a Wizard-of-Oz setup. Ran two conceptual flows — a categorised model vs. a gamified model — with 3 participants in class. The plan was to revise after each session, but the time-intensive Wizard-of-Oz setup meant changes were batched and documented for the high-fidelity build instead.
Usability Testing — Task Completion & Satisfaction Scale
Conducted hi-fi usability testing with participants over Zoom. Expanded the protocol to include a scenario, three separate tasks, post-task 1–7 satisfaction ratings, and open-ended reflections. Replaced SUS with task-specific satisfaction ratings for more targeted, actionable data. Moderators recorded sessions and noted navigation patterns, button interpretation, and overall interaction flow.
Competitive Analysis
Where the market falls short
Five tools were examined across two categories — mainstream photo management and utility-focused cleanup. Across all of them, emotional meaningfulness and personal storytelling were entirely absent as value propositions. Success was framed in terms of gigabytes saved or items deleted, never memories preserved.
Mainstream
Apple Photos
Sets strong expectations: scroll a feed, tap albums, mark favourites, search. Prioritises a single continuous stream and Apple-defined categories over user-defined structure, making sheer volume overwhelming.
Does not help users decide if a photo holds long-term meaning
Mainstream
Google Photos
Works well for basic access and light organisation. Encourages users to buy cloud storage as the primary response to a full camera roll, rather than curating more meaningfully.
Storage-first framing; no sentimental value proposition
Utility
SwipeWipe
Engaging swipe-based model that users are already familiar with. Targets storage reduction — points out near-identical photos and duplicates but doesn't help users decide if any one of them is worth keeping.
Focuses on clutter reduction, not sentimental value
Utility
Clean My Phone
Targets duplicate cleaning and screenshot removal. Shows that users are familiar with high-volume, one-tap cleanup actions and appreciate automation that feels safe and reversible.
No memory-based or emotion-aware curation
Utility
Remo
Similar utility focus to Clean My Phone. Validates the user appetite for automation in cleanup tasks but frames value entirely around freeing up device storage.
Gigabyte-focused; no concept of meaningful curation
Memento's differentiation
While competitors focus on space saving, Memento focuses on meaningfulness — treating photos as reminders of people, places, and stories rather than files to delete. It is designed to be complementary to existing tools: users may still need a duplicate cleaner, but Memento answers the question those tools don't ask: should even one photo of this moment exist in your library?
Research Findings
Four key insights from interviews
Seven semi-structured interviews with contextual inquiry uncovered four interconnected patterns — each one directly informing a design decision in Memento. Participants ranged from ages 25–32 and were recruited via word of mouth.
Curation Is Reactive, Not Intentional
Participants only cleaned up their photos when forced to — by storage warnings, a full camera roll before a concert, or a waiting room gap. Curation happened in short bursts as a secondary background activity (e.g. while watching TV). Some always started from the very beginning of their camera roll, making the task feel repetitive and endless.
Camera Rolls Are Full of Noise
Screenshots, temporary reference images (tickets, dates, recipes, doctor appointment cards), and one-off photos taken to share with others linger long after their usefulness expires. This clutter makes the sheer volume overwhelming before real curation even begins. Participants felt Apple Photos prioritised its own categories over user-defined structure, making retrieval a hassle.
Meaning Is Tied to Stories, People & Events
When describing meaningful photos, participants almost always pointed to images involving people and memorable settings — trips, special events, major life transitions. Meaningfulness was defined by the story behind a photo: who was in it, where it was taken, what the vibe of that time was. Not all pleasant or aesthetic images were considered meaningful in the long run.
Existing Tools Add Friction, Not Guidance
Participants found Apple Maps and search occasionally helpful but inconsistent — effective only when they already remembered a key detail like location. No existing tool helped them determine whether any given photo deserved to stay. Curation was emotionally loaded: a mix of overwhelm and cognitive strain on one hand, and nostalgia, satisfaction, and productivity on the other.
These insights led to two direct design implications: any successful system must help users remove obvious noise before asking them to make harder judgements about meaning; and meaningful photos are best surfaced through their connection to stories, people, and events — not abstract metrics like file size or recency.
"Some photos were good, but don't necessarily mean a good memory."
Usability Testing Participant — on what makes a photo worth keepingData Analysis
Affinity mapping
After listening to recordings and cleaning up notes from all seven interviews, the team transferred key statements and observations into FigJam and grouped them into clusters representing shared themes. These clusters captured patterns that were turned into a written insight document, which fed directly into persona creation and early design decisions.
By visually clustering comments, the team moved from scattered anecdotes to coherent themes that directly informed features — including addressing screenshots and low-value content, and providing ways to lean into stories and emotions rather than pure deletion. The main tradeoff of using notes rather than fully coded transcripts was that very fine-grained or subtle perspectives may have been missed, though the level of detail was sufficient for the project's scope.
User Persona
Designing for Sarah
Insights from the affinity diagramming process were synthesised into a primary persona that guided design decisions throughout the project.
Sarah
Scenario
Gal uses her iPhone every day to capture her life. She takes photos of everything — friends, outings, and little moments she wants to remember. She also snaps pictures of practical things like doctor-appointment cards and event posters so she doesn't lose important information. Over time her camera roll has become huge and scattered across multiple photo locations. She knows she should curate and clean things up, but the idea of going through thousands of images feels overwhelming. Cleaning up photos only becomes a priority when she runs out of storage or starts getting constant "storage almost full" notifications.
Goals
- Get rid of similar, blurry, or otherwise extraneous photos
- Keep only the photos that remain meaningful over a longer period of time
- Reduce the mental effort it takes to manage and find photos
Motivations
- Limited phone storage pushing her to clean up
- Wants to find her favourite photos faster when reminiscing or posting
Behaviours
- Takes daily, casual photos primarily on her iPhone
- Relies solely on phone storage — no external drives
- Uses photos as a quick capture tool for information (appointments, posters, etc.)
- Has images spread across multiple places and apps but is unsure how to organise or migrate them
Pain Points
- Decision fatigue about what to keep versus what to delete
- Overwhelming number of pictures makes it hard to even start curating
- Confusion about different storage and backup options
- Reluctance to pay for extra storage
Design Decisions
What the research drove us to build
Every major design decision in Memento traces directly back to a specific research insight. Four features became the core of the experience.
Batch-Based Review
Users select how many photos to review per session — removing the paralysis of staring at a gallery of thousands. Small, bounded decisions lower cognitive load and make curation feel achievable rather than endless. Literature supported this: structure and playful interaction lower the perceived burden of curation, while tools that feel like admin work get abandoned.
AI-Powered Event & People Grouping
Memento automatically clusters photos by location, time, and people — mirroring how users naturally recall their memories. Noise (screenshots, temporary reference images) is surfaced and handled separately, so harder judgements about meaning come only after obvious clutter has been cleared. Participants found this intuitive and well-matched to their mental models.
Reflective Prompts
Each photo is paired with a contextual question — "Does this photo bring back a good memory?" — framed as a call to reflect rather than a directive to delete. Prompts are framed as questions and the voting buttons as a response, creating a call-to-action model. Participants found them grounding; however, static prompts became invisible over time, pointing to a need for more dynamic prompting in future iterations.
Three-Way Voting: Yup / Not Sure / Nope
Drawing on literature showing voting feels less burdensome than deleting, Memento replaces the binary keep/delete model with three options. "Not Sure" lets users defer difficult decisions and revisit borderline photos — reducing anxiety around irreversible choices. The gamified completion screen (storage freed, progress bar) reinforces a sense of productive momentum.
High-Fidelity Prototype
Key screens
The final prototype was built in Figma across a full curation flow — from the home screen through event selection, photo-by-photo voting, review, album assignment, and the completion state. Below are the nine primary screens.
Homepage
Photo count, detected clutter, progress bar, and three clear entry points. "What mood are you in today?" frames curation as a mood-based choice, not a chore.
Curation Mode
Three paths to curation: AI-grouped by event, AI-grouped by people, or manual gallery selection. Each serves different organisational preferences.
Number Selection
Dropdown to select batch size before starting. Scopes the session and prevents the overwhelm of open-ended review. Usability testing suggested time-based scoping as a future alternative.
Event-Based Review
AI-grouped albums by year, location, and time — each with a cover thumbnail and photo count. Users tap an album to begin curation within that event.
Curation Options Popup
Onboarding overlay that explains all three voting actions and their consequences before curation begins — added after RITE testing revealed confusion about what each vote did.
Curation View
One photo at a time with a reflective prompt above. Three voting options at the bottom. Arrow navigation between photos. Counter shows progress through the batch.
Duplicate Detection
When a near-duplicate is detected, a contextual label ("Detected Duplicate 1 of 2") appears above the photo — helping users make an informed comparison without leaving the flow.
Review & Move Photos
Summary of all three categories, each with thumbnail previews and a description of what happens when the user finishes. Users can navigate into the Yup category to move photos to a named album.
Completion Screen
Gamified summary — photos reviewed, storage freed, and overall progress bar. Option to continue another round or return home. Reinforces a sense of productivity and momentum.
The full interactive prototype is available on Figma. It includes all curation screens, duplicate-detection states, the album-selection overlay, and the complete happy-path flow used in usability testing.
Round 1 Evaluation — RITE Method
Comparing two conceptual flows
Three participants tested two low-fidelity paper prototypes in class using a Wizard-of-Oz setup. Flow 1 leaned toward meaningful decluttering; Flow 2 leaned toward meaningful curation and album building. Both produced distinct insights that directly shaped the high-fidelity prototype.
Flow 1
Categorised / Meaningful Decluttering
What worked
- Two of three users appreciated storage and photo count info on the home screen — made the app's value feel tangible
- Iconography was considered delightful and self-explanatory
- Small prompts above each photo helped users remember why an image mattered
- The "maybe" option was well-received — gave users a way to defer difficult decisions
- Gamified end-of-session summary showing storage freed reinforced a sense of progress
Where users struggled
- Inconsistent icon labels across screens (e.g. "Yes/Meh/Nope" in one place, different wording elsewhere) — one participant explicitly called this out
- "Start Curation," "Quick Recap," and "View Meaningful Moments" were not clearly differentiated — at least one user clicked "View Meaningful Moments" when trying to access the main gallery
- Cleanup modes overlapped — "obvious noise cleanup," "event-based review," and "meaningful curation" all seemed to do similar things; one participant chose noise-cleanup when instructed to focus on memories
Design changes
- Standardise all iconography and adopt action-oriented, consistent labels throughout
- Add lightweight onboarding or info pop-ups explaining what each action does
- Add a thumbnail strip next to the main image while swiping for context
- Simplify home screen to fewer, clearer entry points; make meaningful vs. event-based curation sub-flows rather than competing entry points
Flow 2
Gamified / Meaningful Album Building
What worked
- Favourite button felt intuitive
- Overall system became easier to use over time — the flow is learnable with a short warm-up period
- Automatic grouping of photos (e.g. food, location-based sets) was particularly appreciated — aligned with how users think about events and categories
Where users struggled
- The initial "how many photos do you want to review?" screen confused users; one participant wanted a smaller set than the default, suggesting count-based scoping may not match mental models
- Inside the grouped gallery, two participants were unsure where to click to move forward and couldn't interpret the circle and star icons
- Multiple users were puzzled by why favourited and non-favourited photos appeared in the same group, blurring the distinction between "meaningful" and "regular" images
Design changes
- Separate favourites into a distinct group to match user expectations
- Add clearer on-screen instructions and more visible prompts
- Clearly communicate "what happens next" after a batch is finished — users need to understand the outcome of their actions
- Explore time/date-based scoping as an alternative to count-based selection
Across both flows, all three participants were satisfied with the final set of photos that resulted from the more decluttering-oriented model — suggesting users can feel good about outcomes even when the conceptual framing differs. The sessions gave the team a prioritised list of changes: at highest priority, consistent icons and labels, simpler home-screen navigation, better action explanations, and reduced overlap between modes.
"The prompts were grounding — a reminder. They made me think of old memories."
Usability Testing Participant — on Memento's reflective prompt featureRound 2 Evaluation — Usability Testing
High-fidelity testing results
Participants completed three structured tasks in a standardised scenario — a daytrip to the Art Institute of Chicago with two close friends — using the high-fidelity Figma prototype over Zoom. Each task was followed by a 1–7 satisfaction rating on ease and intuitiveness.
Navigate to meaningfully curate 10 photos from this trip
Two participants clicked "Quick Cleanup" first rather than "Curate Gallery," revealing that the distinction between cleanup modes was not immediately clear — despite added description text below each button. Once users understood the differentiation, the event-based organisation was described as intuitive and well-matched to their mental models: "it was well organised and easy to find." AI grouping was not perceived as intimidating; participants treated it similarly to Apple Photos' location grouping.
Categorise each of the 10 photos as meaningful, not meaningful, or decide later
The lowest average intuitiveness score across all three tasks. Several participants could see how the system might help them manage photos but were not always sure what the feature was doing or how it supported their decisions. The word "meaningful" caused noticeable confusion — participants were unsure whether it referred to emotional importance, photo quality, or something the system was calculating on its own. Reflective prompts were genuinely helpful when noticed, but they became static and invisible after the first few photos. Voting buttons were sometimes described as too small, and participants were unsure what their votes actually changed in the final set.
Move meaningful photos to an existing album and finish the curation process
The highest satisfaction of the three tasks. One participant praised the autonomy the album system gave them. However, some participants did not designate their Yup photos to a specific album as the task required — it was unclear whether this was because they misunderstood the relationship between the review categories and albums, or because the prompt wasn't salient enough. The overall flow felt more natural at this stage, suggesting the interface becomes more learnable the deeper users go into the process.
Cross-task observations
AI usage was not problematic across any of the tasks. Some participants did not register the grouping as "AI" at all — they simply treated it as photo organisation by event, comparable to what Apple Photos already does. When asked about sorting by people, one participant said they would think of it as "similar to Apple Photos."
Participants who noticed the reflective prompts responded to them positively — describing them as grounding, reflective, and evocative of old memories. However, one participant noted that the prompts became unnoticed after a while because they were static. Another participant said they would have been helpful but hadn't noticed them at all until asked. The pattern across participants was clear: the concept of meaningfulness resonated, but the value was not yet fully visible without stronger guidance.
"I could see how this might help me manage and select photos, but I wasn't always sure what the feature was doing or how it supported my decisions."
Usability Testing Participant P2 — on the curation featurePriority Matrix
What we'd fix next
The table below consolidates the highest-priority changes identified across both evaluation rounds, ranked by impact and grounded in direct participant evidence.
| Feature / Change | Priority | Evidence |
|---|---|---|
| Consistent iconography and clear action labels (Keep / Maybe / Delete) | High | Inconsistent terminology caused hesitation and confusion across both rounds |
| Simplify home screen — fewer, clearer entry points with distinct mode descriptions | High | 2 participants clicked "Quick Cleanup" when instructed to curate meaningfully |
| Onboarding pop-up explaining voting actions and outcomes before curation begins | High | Participants unsure what Yup/Not Sure/Nope did to their photo library |
| Clarify grouping logic and remove conceptual overlap between curation modes | High | Event-based, noise-cleanup, and meaningful modes perceived as similar in RITE Round 1 |
| Include delete action in grouped and single-image views | High | Users expected a delete option to be readily available within the curation workflow |
| Dynamic, more visible reflective prompts | High | Prompts useful when noticed; static prompts became invisible after the first few photos |
| Separate favourites into a distinct group from uncurated photos | Medium | Users confused by favourited and non-favourited photos appearing in the same group |
| Larger voting buttons with clearer visual feedback on selection | Medium | Buttons described as too small; unclear whether a vote had registered |
| Improve icon visual design — more expressive while remaining consistent | Medium | Users enjoyed icons overall but noted inconsistency across screens |
| Maintain storage-freed gamification on the completion screen | Low — Keep | Positive reaction during both testing rounds; already well understood |
| Thumbnail strip for comparing similar photos during swiping | Low | Requested during RITE sessions; useful but not essential for MVP |
| Automated grouping for blurry / duplicate / similar photos | Low — Future | Suggested by users; higher implementation complexity, not essential for MVP |
Future Directions
What we recommend
Build Adaptive, Personalised AI Prompting
Static prompts became invisible after the first few photos. Future versions should explore dynamic prompts that change based on image content, context, and the user's evolving curation history — making each prompt feel genuinely relevant rather than formulaic. Users should also be able to fine-tune prompting logic to align the app with their own definition of meaningful.
Strengthen Bulk-Management and Album Organisation
Participants understood the album flow but wanted batch-moving and bulk-management tools that matched the scale of their libraries. More transparent clustering logic — showing users why photos were grouped together — would build trust in the AI and reduce confusion about what "meaningful" means in the system's terms. Long-term photo organisation emerged as an area needing further dedicated testing.
Conduct a Longitudinal Study with Personal Photos
All testing used stock and team member photos due to IRB constraints. The most important next step is an IRB-approved study with participants using their own camera rolls across 4–6 weeks — measuring whether Memento supports lasting behavioural change, reduces cognitive load over time, and encourages ongoing engagement rather than one-time cleanup sessions. This longitudinal dimension is where the most meaningful research still lives.
Limitations & Discussion
What we'd do differently
Looking back across the full project cycle, several constraints shaped the research in ways that future work should address.
Stock Photos Instead of Personal Libraries
Due to IRB constraints and time limitations, participants curated stock photos and team member photos rather than their own camera rolls. A structured scenario helped compensate, but the emotional resonance of personal photo curation could not be fully replicated. Future studies must prioritise this.
Wizard-of-Oz Inconsistencies in Round 1
The paper-prototype Wizard-of-Oz setup introduced variability — different groups sometimes experienced slightly different handling when they strayed off the intended path. With more preparation time, a more controlled and stable low-fidelity prototype would have reduced this variability and allowed users to explore more freely.
Small Sample and Limited Diversity
Convenience sampling via word of mouth limited the age range (25–32) and likely the diversity of photo behaviours uncovered. A participant pool would have broadened the range of mental models and curation habits. RITE sessions used only 3 participants, so findings were treated as qualitative rather than statistically meaningful.
Lack of Quantitative Measures in Round 1
The team would have added quantitative measures earlier in the process — so that Round 1 results could better inform comparisons with Round 2 data. A small concept test before the first usability session could also have anticipated many of the home-screen confusion issues that dominated the redesign.
Despite these constraints, both project goals were achieved. Through user interviews, the team identified users' mental models relating to meaningfulness and found that users are often seeking to curate photos that elicit memories, even if subconsciously. Through design interviews and usability testing, Memento evolved from a concept into a tested, validated interaction model — one that users responded to positively and described as something they could see themselves using.
Photos are not pixels — they are portals
The literature review found that viewing pictures from past experiences elicits sensory memories and emotional detail in a way few other digital artefacts can. Memento's core premise — that curation should be guided by meaning, not megabytes — is grounded in this reality. Every design decision, from the reflective prompts to the three-way vote, was built to honour the emotional weight people already feel about their photos, and to make that weight easier to act on.
Reflection
What I learned
This project reinforced how much is gained by grounding every design decision in research. Our early interviews didn't just validate the idea — they fundamentally shaped it. The insight that meaningfulness is tied to stories and people, not aesthetic quality, was something we could never have reached through competitive analysis alone. It came from listening to people describe their own camera rolls in real time.
I also developed a sharper understanding of how abstract values like "meaningfulness" need to be operationalised for users through concrete interaction design. Participants resonated deeply with the concept but struggled to apply it without guidance — a reminder that even the most compelling idea needs scaffolding to be usable. The confusion around the word "meaningful" in usability testing was one of the most instructive moments of the project: it showed that naming, framing, and in-context definition are design problems, not just copy problems.
Working across both researcher and designer roles in a cross-functional team also sharpened my appreciation for the value of iteration cycles. The RITE sessions felt messy and resource-intensive in the moment — but the changes they produced made the hi-fi prototype measurably clearer. Good research doesn't just answer questions; it tells you which questions to ask next.
Want to see more?
Explore other projects in my portfolio, or get in touch to talk research, collaboration, or anything in between.