iOS product design · Summer 2026

Luma

StatusFigma concept prototype · not yet tested with users

Designing beyond the RSVP.

Luma event feed concept
Choose an event
Text invitation offering an event buddy or attending alone
Arrive with company
Luma reconnection concept
Reconnect afterwards
Problem
Luma gets people to RSVP, but attendees can’t tell if an event fits, often arrive alone, and lose touch with the people they meet.
What I did
On a team of three, we ran 6 interviews and an 8-person survey together. I designed onboarding, the visual system, and the final hi-fi screens for every flow.
Result
A Figma concept prototype of three flows. My onboarding turns 12 answers into a portrait people can review. Not yet tested with users.
Role
Product designerOnboarding, visual system, final hi-fi screens
Team
Three designersChristina, Navneet, and me
Timeline
Summer 2026
Deliverable
Figma concept prototype3 flows · 39 screens
6
interviews
8
survey responses
12
onboarding questions
39
screens across 3 flows

My part

Research
Ran 6 interviews and an 8-person survey with the team
Onboarding
The 12-question preference survey and the portrait at the end
Visual system + hi-fi
One visual language, and the final hi-fi screens for all three flows

We designed as one team. Christina led Event Buddy and secondary research; Navneet led Plans, post-event messaging, and competitive analysis.

Problem

An RSVP is not a connection.

Luma handles discovery and sign-up. It stops before the parts attendees worry about: whether an event fits, who to arrive with, and how to see people again.

Original Luma Discover screen
Discover · original
Original Luma event detail screen
Event detail · original
51%
of Americans find it difficult to make new friends
62%
say making friends was easier at another time in their life

Pennington, Hall & Holmstrom, PLOS ONE, 2024: national surveys of U.S. adults, 2022–2023.

  1. Choose

    Deciding is hard.

    A Luma reviewer asked to save events while still undecided. App Store, Aug 2025

  2. Arrive

    Arriving alone is a worry.

    One Luma host already offers to pair people who don’t want to come alone. Host’s event page

  3. Reconnect

    Following up feels awkward.

    An attendee wasn’t sure how to follow up with people they met. Reddit, Jul 2025

Public examples, not study participants. The Reddit post is about Meetup, a similar event app. Luma already has chat; this concept adds guidance around it.

Research

6 interviews, 8 survey responses.

We ran 6 interviews and collected 8 survey responses. The counts below come from the survey; they describe this small sample, not event attendees generally.

  1. Company

    4 of 8 wanted a buddy.

    Make pairing optional.

    The other four were fine either way.

    Led toAn opt-in Event BuddySee the decision
  2. Atmosphere

    6 of 8 chose structured activities.

    Show what an event feels like.

    Six respondents selected structured activities, informing the decision to ask about event format during onboarding.

    Led toPreference onboardingSee the decision
  3. Comfort + access

    7 of 8 picked an identity or access option.

    Ask about comfort directly.

    This count combines identity and access options. It does not identify which specific accommodation each person needed.

    Not yetNothing yet

Full research synthesis in Figma.

Design decisions

Three decisions, each from a finding.

Onboarding · designed by me

Ask what the room should feel like.

Evidence
6 of 8 survey respondents chose structured activities. Event format became an onboarding input.
Considered instead
Progressive onboarding, asking a few questions at a time.
Decision
Ask about vibe, size, and setting, then show a preference portrait. The sketch, wireframe, and built screen below show how that decision took shape.
Preference portrait after twelve onboarding questions
The payoff · review your answers

How it got here

Sketch listing twelve onboarding questions in order
Grayscale wireframe of the vibe question
Built vibe question with tinted selected states

Still to testThe survey supports asking about format; it does not establish that twelve questions are the right length. Test completion and whether the portrait feels worth the effort.

What each answer changes

  1. Q1Buddy, solo, or eitherWhether Luma offers you an Event Buddy before the event
  2. Q2Event sizeSmall / Intimate or Large / Lively tags; reconnection suggests small events you’d both enjoy
  3. Q3Indoor or outdoorIndoor / Outdoor tags; shown as a shared trait when choosing a buddy
  4. Q4Drink preferenceAlcohol-free tag on events; kept private, never shown to a match
  5. Q5VibeHigh-energy, Low-key, Mingling, Structured tags
  6. Q6SeatingSeating tag
  7. Q7ActivitiesTopic tags: Tech, Food & Drink, AI, Arts & Culture, and more; a buddy match reason: “You both like food events”
  8. Q8TimingWeekday / Weekend and Daytime / Evenings tags
  9. Q9DistancePartly: city and district tags, but no distance filter
  10. Q10How oftenNothing yet
  11. Q11Who you go withNothing yet
  12. Q12CostNothing yet
Eight answers change something you can see. Three don’t yet, which is the case for cutting them or designing their payoff first.

Event Buddy

Offer company. Keep the choice.

Because the survey split 4 to 4, the invitation gives two equal answers: find me a buddy, or go on my own. A notification starts it, and the reply happens over SMS.

Luma Event Buddy notification on an iPhone lock screen
Notification · an entry outside the app
Text invitation offering a buddy or attending alone
SMS · two equal choices

What I learnedWhen research splits evenly, design the choice instead of picking a side.

  1. V1 · Word replies

    Reply “Yes” or “No”

    What I built
    The first text asked people to answer in words.
    Why it failed
    The team’s note on the flow: “assume that people will misspell.” A typo leaves the reply unreadable.
  2. V2 · Final

    Reply 1 or 2

    What I built
    1 for “Yes, find me a buddy,” 2 for “No, I’ll go on my own.”
    Why it works
    One digit is hard to mistype, and the two options stay equal.

Reconnection

Put the people you met next to your plans.

From Plans, find someone you met and choose a shared event or a message.

Considered insteadA separate Connections tab, as in the planned five-tab bar.

TradeoffConnections moved into Chat and Plans. One fewer tab, but a less direct route to a person.

Reconnection prototype · from notification to the next conversation
Planned tabs
HomeDiscoverPlansConnectionsProfile
Built tabs
HomeDiscoverPlansChat

What I learnedFewer tabs looked cleaner, and made the key task harder: finding someone you met.

All three decisions in one prototype. Type your name, answer the twelve questions, then tap “Explore my feed” and keep going into Event Buddy, Plans, and Reconnect. Onboarding is coded; the other flows are the final Figma screens with their prototype links.

Visual system

One visual language across three flows.

Luma visual system board with color, type, spacing, radii, and UI states
I rebuilt Luma’s shared patterns from reference screens, then used the same type, spacing, and states across all three flows.

What I learned

The payoff came too late.

No launch metrics here; it’s a concept. What I’d carry into the next project:

Next user test · proposed

  1. OnboardingSet preferences and choose an eventObserve hesitation or abandonment; ask what answering helped them do
  2. ChoiceAttend without a buddyCheck whether the solo path is discoverable without assistance
  3. ReconnectFind someone you met and arrange a next stepObserve the first tab chosen and where the person gets stuck
Planned tasks, not test results.
  1. I cut the question my research said mattered most.

    7 of 8 people picked an identity or access option, and I replaced that question with “who do you go with.” Next time, every cut gets checked against the findings, with the reason written down before a question leaves the flow.

  2. I built the input before the payoff.

    I prototyped twelve questions and a portrait, but not the “why this event fits you” screen that makes answering worthwhile. Now I design the payoff screen first and work back to the fewest questions it needs.

What’s next

Not built yet

  • “Why it fits” explanations
  • The identity question
  • An editable Profile
  • A Connections view

Test next

  • Onboarding length
  • Whether people can find someone they met
한국어