☕️Grindtime
ROLE
Full-stack designer
TOOLS
Figma, Otter.ai, Goodnotes, Adobe Illustrator, Whiteboard
TIMELINE
2 weeks (December 2025, or winter break)
This project was independently designed and developed as a personal project. The idea was inspired by by firsthand experience navigating café workspaces in New York City.
Table of Contents
1. Context
2. Problem
3. Design
4. Prototype
5. Next Steps
6. Conclusion
1. Context
Residents of NYC face a unique problem: Where do I work/study?
Let's take your typical NYU college student -- Emily.
User Personas
If Emily wants to find a quiet place to study or do some work, her only options are:
-
Her dorms/apartments?
-
She's spent all night there, and wants a change of scenery.
-
-
School libraries?
-
Students in most other cities/college towns have bigger libraries or campus centers. That's harder to find in the city, where space is constrained.
-
-
City libraries?
-
City libraries tend to be crowded. The ideal time to go is in the day, but students have class then. And in the afternoons and weekends – families and young children abound.
-
-
Campus centers?
-
NYU is small, and space is constrained.
-
-
Coffee shops?
-
Could work, but finding one is hard.
-
Note: We also considered a second user persona -- the WFH professional. While the primary user research for this project focused on college students, this second persona emerged through contextual observation and pattern recognition rather than direct interviews. Due to access constraints, this group was not studied as deeply, but their needs closely parallel those of students in terms of decision uncertainty and execution challenges.
Why Existing Tools Don’t Work:
Across these tools:
-
work-relevant attributes (noise, seating, outlets) are implicit or buried,
-
users must mentally simulate how a space might feel,
-
Results are delivered as long lists, increasing decision fatigue.
2. Problem
Emily's story is a synthesis of 15 people with whom I conducted interviews and sent a survey to. This sample of 15 college students are between the ages of 19-21 and have completed at least a semester of study in NYC.
Frameworks applied
“5 Whys” Root-Cause Probing
I applied the Five Whys framework to move beyond surface-level complaints (e.g., “it’s hard to find a café”) and uncover the underlying behavioral and cognitive drivers behind café selection. This approach aligns with first-principles reasoning and was used during informal interviews and follow-up questions to strip stated preferences down to root causes.
Examples from interviews:
When one respondent said, “I usually just use Google Maps to find a coffee shop,” follow-up questions revealed deeper motivations:
-
Why Google Maps? → It’s the fastest way to see what’s nearby.
Why does that matter? → Because time windows to work are limited.
Why is that important? → A bad choice wastes time and disrupts focus.
Similarly, when another user said, “Many places don’t satisfy all three,” further probing revealed that the core issue wasn’t discovery, but uncertainty around whether a space would actually support working (seating, Wi-Fi, laptop policy).
This iterative “peel back the onion” process reframed the problem from “finding cafés” to reducing decision risk and cognitive overhead under time pressure.
Jobs to be done (JTBD) framework
I used the Jobs To Be Done (JTBD) framework to focus on the underlying outcome users were trying to achieve, rather than their stated tool preferences or demographics.
Across interviews, users referenced tools like Google Maps, Yelp, TikTok, and subscriptions (e.g., Pret). These were reframed not as solutions, but as workarounds for a deeper job.
Reframed job:
“Help me confidently choose a place where I can focus for this specific work session — and actually follow through once I arrive.”
For example:
-
A student prioritizing “seating, Wi-Fi, and outlets” was not asking for more filters, but for confidence that a choice wouldn’t fail.
-
A user defaulting to Pret due to a subscription was not optimizing for brand loyalty, but for predictability and low-risk execution.
This JTBD lens guided synthesis away from feature-level requests (lists, ratings, reviews) toward designing a decision-and-commitment system that supports both selection and focused execution.
Key takeaways
73%
of interviewees said they have arrived at a café once, only to realize it didn't have a wi-fi or outlet.
“The difficult part was just finding a place with seats, fast Wi-Fi, and that let you work there — many places don’t satisfy all three.”
Respondent #2 (student, Duke in New York)
“I did go to a lot of Prets, mostly because I had that monthly Pret subscription (for my daily coffee run ) and there’s Prets like everywhere and they almost all have wifi."
Respondent #8 (student, Duke in New York)
"I don’t want to haul all the way there for nothing.”
Respondent #5 (student, NYU)
Customer Journey
Based on our conversations, we mapped out the typical cafe-work process:
KEY TAKEAWAYS
My research lead me to identify a few key patterns that guided my design process.
1.
The “Three-Requirement Problem”
Most users implicitly require all three:
-
seating,
-
fast Wi-Fi,
-
laptop-friendly policy (this includes outlets)
Many cafés satisfy one or two — but missing any one breaks the session.
2.
Decision-Making Is Risk-Averse
Users want to explore new places, but default to Google Maps, libraries, Pret/chains, because experimentation has a high failure cost.
3.
Economic & Habit Constraints Matter
Subscriptions (e.g. Pret) and familiarity act as anchors, not preferences:
-
“I know it has Wi-Fi.”
-
“I can save money here (Pret membership, Starbucks stars)"
-
“It’s everywhere.”
3. Design
Lo-Fi Wireframe
Lo-Fi Wireframe
Design Rationale
My goals with the visual components were to reduce cognitive load, encourage calm decision-making, and support sustained focus.
A warm, neutral color palette was chosen to evoke the comfort of a café environment without overstimulation. Soft beige and cream tones create a low-contrast backdrop that minimizes visual noise, helping users feel settled rather than distracted while making decisions or working through a session.
Satoshi was selected as the primary typeface for its balance of clarity and warmth. Its clean geometry -- especially for a Sans Serif font -- ensures readability at small sizes, while its subtle personality avoids the coldness of purely utilitarian fonts—supporting both focus and approachability
Subtle glassmorphism and elevation cues are used to establish hierarchy without hard borders. To me, this was one of the key design trends of 2026 for a reason -- it allows key actions and cards to feel distinct while maintaining a sense of visual continuity, reinforcing the idea of a single, coherent flow rather than fragmented steps.
The subtle fade-in transition eases users into the decision-making flow, reducing cognitive load at the moment they’re asked to articulate constraints
Light kinetic interactions (hover states, transitions, and micro-movements) provide feedback and momentum without drawing attention away from the task.
An interactive cross-off feature and the confetti-animated pop-up were meant to create a design-your-own reward system.
4. Prototype
5. Next Steps
Backend Development
With the core interaction and user flow validated, the next phase of this project would focus on strengthening the system behind the experience: building the backend infrastructure required to support reliable recommendations and creating feedback loops that allow the product to improve over time.
Data Pipeline
To move beyond a static or manually curated experience, the product would require a backend that aggregates, normalizes, and serves café data in a scalable way. Early development would intentionally combine structured personal data with external sources to balance quality and coverage.
1. Café data ingestion
-
Seed the system using a manually curated spreadsheet of NYC cafés that I’ve maintained over time, containing locations I’ve personally vetted or researched. This spreadsheet is an accumulation of research I've done since I began tracking cafes in my sophomore spring (Spring 2024)
-
Supplement with external sources (e.g. Google Places) to expand coverage and fill gaps
-
Allow future ingestion from user submissions as the dataset grows
2. Attribute normalization
-
Standardize work-relevant attributes such as Wi-Fi reliability, outlet availability, seating comfort, noise level, and typical stay duration
-
Translate qualitative notes (e.g. “laptop-friendly,” “busy after 3pm”) into structured signal
Feedback Loop & Learning System
To continuously improve recommendation quality, the system should learn from real-world outcomes after a session concludes.
1. Lightweight post-session feedback
-
One- or two-tap prompts (e.g. “Did this café work for you?”)
-
Optional and low-friction to avoid disrupting focus
2. Adaptive recommendation tuning
-
Use feedback to update café attribute confidence scores
-
Identify recurring mismatches (e.g. cafés that fail longer focus sessions)
3. Internal insight generation
-
Surface trends around noise, session duration, and location
-
Inform data cleanup, prioritization, and future feature decisions
🌟 NORTH STAR METRIC
café selection rate
=
% of search sessions that result in a café being selected without additional searching
Why this metric:
-
This metric directly measures whether your product is doing its core job: reducing decision fatigue, increasing confidence under uncertainty, helping users commit to a choice quickly
-
It captures quality of reasoning, not just clicks, unlike: “Searches per user” (vanity), “Café views” (exploration-heavy),
or “Time on page” (ambiguous). -
This metric instead answers: Did the system help the user decide?