Infinight — A Product Story

Every Saturday, a plan dies in the group chat. Infinight brings it home.

A going-out app that helps a group pick one night, book it together, and get everyone home safely, from the first “what are we doing?” to the ride back.

Independent research & design project · Bengaluru · 2026

Checked in Who's here Rate the night

The solution, at a glance

One app for the whole night, from "what are we doing?" to the ride home.

Pick a mood, get three places

Tap how tonight feels. Infinight comes back with three places, each with one reason it fits.

Let the group vote

Share one link. Friends vote on the three options and the plan settles itself.

Get everyone there and home

Rides are booked with the table, and the app checks everyone got home.

Start with the person

Going out should be the easy part of the week. Right now it's the hardest.

Four friends want the same thing: a good night. What they get is an hour of links, budgets that don't match, a worry about getting back at 1am, and a bill nobody wants to bring up. The night out isn't hard. Deciding it together is.

so what are we doing tonight?

Deciding together. Three to five people, one thread, no shared view of the options.

is it safe there late?

Feeling safe. Safety is researched by hand, if at all, and mostly by women in the group.

reached home?

Getting back. Rides, check-ins and the split are left to whoever is still awake.

…and that gap is a business opportunity.

The pitch was simple: build an app that actually decides with you, instead of handing you a longer list to argue over — fewer filters, less scrolling, just a plan the whole group can say yes to before the night talks itself out of happening.

01

Ticketing apps sell the seat, not the decision

BookMyShow, District and Swiggy Scenes compete on inventory. Whoever owns the moment a group agrees owns the booking that follows.

02

One group, four bookings

A squad plan turns one ticket into seats, rides out and rides home for everyone, each a separate transaction.

03

Revenue past checkout

Booking fees, ride-partner commission, and venues paying to reach the right crowd on the right night.

What Infinight offers

together

Decide together

Mood, budget and the group chat get folded into one suggestion — nobody sets filters alone.

faster

Faster to a plan

Four taps and three picks replace six apps and forty unread messages.

safer

Safety, on the surface

Safe-verified venues and live location sit on the card itself, not three menus deep.

The city is full of things to do. Choosing one is the problem.

Try it yourself

Tap through the real thing.

This phone is live. Start on Home, pick a mood, and follow the night through to the ride back.

What already existed

Everyone helps you find. Nobody helps you decide.

Before talking to a single person, I scanned what Indian users already go out with. Each app is good at one slice of the night, and each one is built to pull a different kind of person in. Tap through them.

{{ comp.nm }}

{{ comp.kind }}
What it offers

{{ comp.offers }}

Behaviour it attracts

{{ comp.attracts }}

Hooks that pull users in

{{ comp.hooks }}

Where it stops

{{ comp.stops }}

None of them know the other five exist, and none of them stick around past checkout. The decision, and the ride home, belong to nobody.

My first guess

Before I asked anyone, I made a bet.

I'd already lived the six-apps-for-one-Saturday problem myself, so before I talked to a single person, I looked at what was already out there — Instagram, Google, BookMyShow, Uber, WhatsApp — and noticed the same gap in all of them: they help you look, but none of them help you actually decide. That was enough to write a guess down.

“Give a group one clear plan they can all just say yes to, and they'll stop arguing about where to go.”

My guess — I hadn't tested it yet.

Talking to real people

I went in with questions, not a pitch.

I wasn't testing an idea yet — I didn't have one. I just asked people to describe how a night out actually gets decided, and let them talk. Most of what I learned came from the twenty minutes right before a plan got made, not from asking about it afterwards.

What I actually did

—I asked people to just walk me through their last night out, start to finish, in their own words
—I asked open, non-leading questions — no feature ideas, no app in mind, nothing to react to
—I read real group chats end to end, from the first "what are we doing?" to the final Uber split
—I asked the person who always ends up being the organiser what that actually felt like
1

It's never one person deciding

Almost every plan I looked at got argued out by three to five people in a chat. Designing for one person choosing alone would have been solving the wrong problem.

2

Nobody talks in categories

Nobody typed a venue name or a filter. The actual messages read like ‘I just want to chill’ or ‘let's go somewhere like that place we went to.’ People describe a feeling, not a set of criteria.

3

"My friend went there" beats any review

The single biggest reason someone picked a place was one friend's name attached to it. Get that wrong and it's creepy. Get it right and it's the best feature in the app.

4

The night doesn't stop at the ticket

The ride home, splitting the bill, finding the car — that's where people actually trust an app, or stop trusting it. Almost nobody bothers designing those screens.

Where the idea actually came from

It wasn't my idea first. It was in their chats.

I wasn't looking for this. Reading through actual group chats, the same kind of line kept showing up, from different people who'd never met each other. Nobody was giving me a list of criteria. They were describing a feeling, to another person, in plain language — and then someone had to translate that feeling into an actual filter search on a completely different app. What if the app started from the feeling — a mood, who's coming, a budget, how far — and came back with somewhere real?

That's the moment this stopped being a filter problem and became a mood problem.

Saturday squad · 4 people
PriyaI just want to chill.
RohanLet's go somewhere like that place we went to.
AishaSomewhere nice, not too far.
😌 ChillSquad of 415 min
↑ the same three lines, as taps

Was I right?

Part of it held up. Part of it, I had wrong.

✓ This part held up

The stalemate is real — I watched plan after plan quietly die in the group chat, exactly where I'd guessed it would. People really do want one clear answer instead of a longer list.

↻ This part I had to change

Two things. First, I'd written my guess for one person — real groups told me it has to work for three to five people arguing at once, not one. Second, I assumed the fix would look like a better filter screen. It doesn't — people don't describe what they want in criteria at all.

My guess, corrected: let a whole group just say what they want, in their own words, and get back one plan they can all agree on.

Everything I saw go wrong

Six problems kept showing up, every weekend.

{{ p.n }}

{{ p.h }}

{{ p.p }}

← the one I chose

The problem I chose to solve

Of the six, social friction — the group chat that kills the plan — was the one I picked: the most frequent, the most emotionally charged, and the reason the whole project started.

Karl · 27 · the organiser

Same thing happens to Karl, most Saturdays.

He's the one who always says "let's actually do something tonight" — then spends the next forty minutes losing a group-chat argument about where. By the time everyone agrees, half of them would rather just stay in.

Ends up being frustrated..

With six apps open and nobody actually deciding, picking the right night out becomes its own chore — the group either burns an hour on it, or defaults to the same bar again.

Before any of this existed

The same Saturday, before I built anything.

{{ s.time }}
{{ s.t }}{{ s.mood }}

None of this is a missing feature. It's people getting stuck deciding together — and that's the one problem, out of the six, I chose to actually fix.

How I got here

Four questions that shaped every decision.

1Listened

What actually happens before anyone leaves the house?

I sat in on real group chats and followed the whole thing, from "what are we doing" to the final Uber split.

2Focused

Which part of the night breaks first?

The group chat. Plans die there before they start, so that became the one thing to fix. Everything else waited.

3Tested

What would make the group decide faster?

I tried a lot of ideas and threw most of them out.

Five-step quizLive map of real peopleStreaksStar ratingsAuto-split with selfiesFour-tap mood pickerGroup pollOne shared tab
4Built

What does the first real version need?

A four-tap mood picker, a group poll, and one shared tab for the night.

Early process · the whiteboard

Started sketching some ideas and mapped them out.

Tap a colour to see one kind of note across the whole journey. ★ marks the ideas that shipped.

{{ col.title }}
{{ n.t }}★

Crazy eights

Eight sketches in eight minutes. Five didn't survive the first real conversation.

Q1 · what music?Q5 · budget?

Five-step onboarding quiz

✗ Too long before you see anything

NehaArjun

Live map of real people

✗ Felt creepy in testing

12week streak

Going-out streaks

✗ Pressure, not fun

Star ratings for venues

✗ Reviews already exist elsewhere

Auto-split with selfies

✗ Too much friction at 1am

show three

Mood chips → three picks

✓ Fast, one tap, keep

Skye · 3 votes

Three-option squad poll

✓ Ends the group-chat loop

Ticket · ₹2,400+ ride home 1am?

Ride home inside checkout

✓ Solves the ride home

From sticky note to screen

Three starred notes, and where each one ended up.

Pre-select one mood chip by default so home is never empty
↓

Home opens on “Tonight feels like…”. Nine moods, one tap, never a blank box.

Poll closes on a soft deadline, never a hard lock-out
↓

“Closes 7:30pm” is a nudge on the squad's plan, not a gate.

Schedule the ride home at the same time as the table
↓

“Book a ride back home?” sits inside checkout, before payment.

Got a good starting point — but it needed real conversations with real groups before any of it was worth building.

The best nights start with a feeling, not a filter.

01

Deep dive 01 · Picking the place

How Infinight decides with you, in four taps.

It starts on Home. The mood chips under “Tonight feels like…” aren't filters; each one opens Shape tonight with that mood already picked. Three more taps — who's coming, spend, how far — and you get three places back.

STEP 1 · HOMETap a mood — say “Chill & Laid Back”
→1 tap
STEP 2 · SHAPE TONIGHTMood is set. Pick who, spend and distance
→3 taps
STEP 3 · TONIGHT'S THREEThree places, one reason each

These are live — tap a mood on Home to follow the flow yourself.

Up close · the decisions inside the screen

01
You answer like you'd tell a friend

Tonight feels, who's coming, spend each, how far. Four questions, each a row of chips — no filter screen to set up first.

02
Context it counts for you

Light rain, it's Saturday, places you've liked — listed above the button, so nothing about the pick feels like a black box.

03
Three places, not thirty

The answer is always three, and the “Why:” line sits on top of each card. The full list is one tap away.

Assumptions behind this flow

The flow assumes the user already has a few interests saved from onboarding.

Partly earned, partly not — onboarding was cut from a five-step quiz to three taps (less cold-start data), so the picks lean on session signals to cover the gap.

Assumes the app has live pricing and availability for the venues it suggests.

Not verified against a real venue API — a product requirement the prototype stands in for, not something the research tested.

Group context (who's coming, past nights) is treated as one signal, not a separate step.

Grounded in what research showed — almost every plan was made in a group chat of 3–5 people, so a single-user flow would be solving the wrong problem.
STILL TORN?

Compare the top two, by the one thing that differs

Instead of a longer list, the compare screen leads with the single difference that matters — quiet till 10 versus loud from the start — and ends in two actions: pick it, or ask the squad.

Once you pick a place, sharing it with the group is one tap away.

Deep dive 02 · The squad

Turning a group chat into a plan everyone picks.

Assumes the group has already agreed to go out — this poll is about where, not whether.

Every session watched started after that decision was already made; the argument was always where, never whether.

Assumes everyone can open the plan from a shared link, even without the app installed yet.

Direct response to a friction point in testing — nobody remembered usernames, so an install-first invite would have killed the group before it formed.

Bill splitting is treated as a later step, once the venue is actually locked in.

Matches the study's own sequence: splitting happened right at the end, from the first "what are we doing?" to the final Uber split — never before.

{{ sqCur.label }}

{{ c.n }}
{{ c.t }}

{{ c.d }}

Once the plan is locked, Infinight books the table and moves on to transport.

Deep dive 03 · There and back

Getting there — and getting home — before you leave the house.

Location access is requested the moment the plan is made, through the normal OS permission prompt — that's what powers the travel time and safe-route data.

A real, working mechanism, not a guess. The one gap: if someone declines the prompt, that fallback isn't designed yet.

Ride and metro prices come from a live feed from each provider, refreshed every time the screen loads — never a stored estimate from last week.

That's the requirement this screen is built against. In this prototype it's mocked data standing in for that real integration, not tested against it yet.

The return trip is only scheduled if the user opts in at the time of booking.

A deliberate call, not a finding — the study showed nights get improvised, so nothing about "going home" gets locked in without asking first.

Neither the traffic forecast nor squad-wide cab booking came out of the research — they're a direct answer to "the 2am problem" and "I want to know the room before I'm already in the Uber," but untested with a real group so far.

{{ trCur.label }}

{{ c.n }}
{{ c.t }}

{{ c.d }}

While travelling to the destination

Detected automatically — no check-in needed
→
At the door
→
Checked in and settled

After checking in, the app goes quiet on purpose — until the recap the next morning.

A feature I almost cut, then brought back

The map isn't a filter. It's a mood ring for the city.

v1

A live "who's here now" heatmap of actual people — I killed that version outright.

v2

What survived is the useful half: aggregate crowd energy by area, with nobody's location exposed. It runs underneath discovery, squad planning and transport.

Heat, not a filter

A cluster tells you where the city is actually moving tonight, before you pick a direction.

Areas, never people

No names, no dots for individuals — just how busy it is right now.

Tap it, and it tells you why

A set just started, three venues are near capacity.

Coarser and private beat precise and creepy — that trade-off was the whole point of bringing it back.

Per-venue crowd, through the night

Who the night is for

One app, but not one kind of night out.

The biggest mistake in my first draft was designing only for my own Saturdays — a group of friends who already knew each other. Solo and safety-first needs weren't something I set out to study — they kept coming up sideways, inside the same group sessions. What's built here is a bet based on what surfaced unprompted, not the same amount of dedicated research the core flow got.

{{ a.who }}

{{ a.h }}

{{ a.give }}
{{ audCur.who }}

{{ audCur.h }}

{{ audCur.p }}

What it gets · {{ audCur.give }}

How the app actually knows

A quick question, not a locked-in identity

{{ r.tag }}

{{ r.t }}

Two ways in

Shaping tonight is the default. It's not the only door.

Most nights, people arrive with a feeling, not a query — so the mood chips are the resting state. But some nights you know exactly what you want, and being walked through four questions is its own kind of friction. So there are two ways in, one tap apart.

Door 1 · Shaping tonight

Four taps, three places

Tap a mood on Home, then who's coming, a budget and a distance. You get three places back, each with one reason.

Insight

Nobody in research typed a venue name to start — they typed a mood. Mood chips match how people actually begin.

Decision

Mood chips are the default surface; they sit on Home, not in a menu.

Door 2 · Searching instead

One tap to the plain list

Explore is a traditional search and filtered list — type a venue, an artist, a category. The filters you'd already set carry across.

Insight

Someone who knows the exact rooftop they're headed to doesn't need four questions — they need a search box.

Research

Repeat and power users wanted a fast list they could scan; forcing the mood picker every time annoyed them by the third visit.

Trade-off

Two ways in cost more to maintain and risk a confusing default. Worth it — keeping search one tab away defuses most of that.

The screen that matters most

One card carries the whole decision.

If a person only ever sees one screen, it's this one. So I pulled every question I watched people ask — is it good tonight, is it safe, will we fit, what's the food, has anyone I know been — onto a single card, and fought over the order of it for weeks. Here's every element, and why it earned its place.

1342
1 · The reason, above everything
3 · Women safety checked badge
4 · How full it is, right now
2 · Real, time-stamped photos
Top of the card
5678
5 · Seats available
6 · Parking & restroom, up front
7 · A peek at the food menu
8 · “Priya went here”
Scrolled down
1

The one-line reason, above everything

The “why” sits above the photo and far above the price. It was the change testers noticed most.

2

Real photos of the actual room

Time-stamped venue photos from tonight, never a map thumbnail or a brand banner.

3

Women safety checked badge

A specific badge on the photo itself, where women-led groups look first.

4

How full it is, right now

Live crowd level. Where the data is thin, the card says so instead of faking a number.

5

Seats available

A plain stat up front, so a group knows it fits before tapping Reserve.

6

Parking & restroom, up front

The two questions people texted friends before leaving, answered next to seats.

7

A peek at the food menu

Two or three signature dishes inline; the full menu one tap away.

8

“Priya went here”

Only friends who chose to share, shown as one warm line. Never a list of strangers.

The trade-off: the reason line takes vertical space I wanted for photos. It won every time in testing.

Brand, style & components

An owl and a map pin: a guide for the night.

The mark combines two ideas. The owl is a creature of the night; the location pin is a real place to go. Together they describe what the app does: find somewhere to be tonight, and keep watch until everyone gets home.

OWL ——

Owls come alive after dark. It stands for the hours the app is built for.

PIN ——

The body is a map pin. Every plan ends at a real place.

Infinight owl-pin logo
—— EYES

Wide open, keeping watch over the group until everyone is home.

—— TOGETHER

Owl and pin as one mark: a guide for the night out, from where to go to getting back safe.

Ember-brow logoPrimary · ember brows
White-brow logoOn purple · white brows
Infinight lockupInfinightLockup
Scales to 22px

Colour, as used in the app

#0B0812Night · app background
#3A0F57Plum · Home header
#8B1FD6 → #B429C8Primary actions
#C77DFFAccent · the “why”
#130E1ESurface · cards
#F3EDFAText · selected chip
#43E08ASafe · home
#FFC44DCrowd · filling up
#4DE3F0Live · rides
#FF5470SOS · alerts
#E2571DEmber · logo only

The ember brows are the logo's one warm accent and stay on the logo; the app itself runs on night purples, with green, amber, cyan and red kept for live status.

Aa AaNewsreader · story headlines
AaBricolage Grotesque · app UI & body
😌 Chill🎵 Live MusicShow tonight's three

The rest of the build

Four complete flows, screen by screen.

The deep-dives so far cover the one problem this project is actually about — picking a place together, the squad, getting there and back. These four flows are the rest of the app end to end. Every screen below is the live prototype.

01

Onboarding to booking

From the first launch to a paid ticket
Launch
Onboarding
Pick a few interests
Home, mood-first
Event details
Review & pay
You're in
02

Sharing the ticket with friends

One tap from the confirmation screen, no account required to open it

Share the plan
→
What a friend sees
03

Booking the ride home

For one person, or the whole squad, from the same screen

Ride back, booked
→
Rides confirmed
04

Reviewing, then picking the next night

The loop that brings someone back
Split the night
→
The morning recap
→
Rate the night
→

How it compares

Other apps own a slice of the night. Infinight owns the whole of it.

The night →
{{ h }}
Stages
{{ m.nm }}
{{ m.count }}/7
Covers itPartlyNot at all

Only on Infinight

Three picks for the whole group

Mood, who's coming, budget and distance become three places with one reason each, ready to send as a poll.

The ride home is part of the booking

The return ride is offered at checkout, before you pay — not remembered at 1am.

One tap, separate payments

The organiser books cabs for everyone; each rider pays for their own.

Everyone home, without asking

Home is a checkpoint. The status turns green when each phone gets back.

Safety and crowd, on the card

Women safety checked and how full it is right now, sitting on the photo itself.

The night after, handled

An even split by default, a recap in the morning, and a rating that shapes the next pick.

The same Saturday, rebuilt

Here's what changes once it's built.

{{ s.time }}
{{ s.t }}

Same bar. Second argument. Nobody home on time.
8:00PM

It's 8pm and the group chat has gone quiet — nobody wants to be the one who suggests where.

…
8:02PM

Karl opens Infinight and taps a mood chip instead of typing a query.

8:03PM

Infinight weighs mood, budget and the group's history, and comes back with three places — Skye on top, with a one-line reason why.

8:05PM

He shares the plan as a poll: three options, a soft deadline. The group votes in one shared thread — no switching to WhatsApp to argue it out.

8:30PM · PLAN LOCKED

Once it locks, Infinight books the table and quietly starts working on transport.

8:31PM · RIDES SORTED

Getting there and getting home are both planned before anyone leaves the house.

NEXT MORNING

At the venue the app checks everyone in, then goes quiet until the recap the next morning.
02

One story from the middle of the project

During round three of testing, I hit a problem with the event card.

After watching people scroll straight past the reason, I realised the card read like a listing — not like a friend telling you why.

Before · didn't work
why: quiet till 10…
1
2
1

Why does the photo lead? It doesn't tell you if tonight's actually the night to go.

2

The price sits alone at the bottom, with no reason attached to it — easy to scroll straight past the "why".

After · in the prototype
1
1

The reason now leads — “Why this, tonight” sits above the photo, far above the price. It's the first thing anyone reads.

It worked in later testing — trust went up, and so did the taps.

User testing

Two things I got wrong, and one I almost cut.

Got it wrong
Explore
Search venues
RooftopLive musicQuiet
"opens a new page?"
Filter chips looked like navigation, not switches

Chips like "Rooftop" and "Live music" sat side by side, styled like tabs. People tapped one expecting it to open a new screen — then got confused when nothing moved. The fix: filling the chip solid the instant it's on, so tapping it visibly changes state instead of going anywhere.

Almost cut it
Last nightSat · The Humming Tree
You spent₹1,850
Favourite moment
The setRooftopLate fries
I nearly dropped the morning-after recap screen

It felt like scope creep this late — photos, what you spent, your favourite moment from the night. But when I asked testers why they'd open the app again days later, this was the screen they brought up unprompted. I kept it.

The card, redesigned three times

The third mistake was the biggest one: the one-line reason started out at the bottom of the card, under the price. Testers scrolled straight past it every time.

v1 · Marketplace cardCut
Tonight
The Humming TreeIndiranagar · 2.1 km · ★ 4.3₹1,200 for two
SaveBook

Read like a listing. Nobody looked twice.

v2 · Reason at the bottomCut
Tonight
The Humming TreeIndiranagar · 2.1 km · ★ 4.3₹1,200 for two
Quiet till 10, live jazz after
SaveBook

Testers scrolled straight past the point.

v3 · Reason on topShipped

Trust rose. Taps followed. This one shipped.

Moving the reason to the top is what made people actually trust it.

Did it actually work?

Does this really end the stalemate?

For the main problem — I think so. Testing backed up the guess I corrected earlier. Everything I built on top of it since, I'm only as sure about as the evidence I actually have.

Still a betBacked by testing

Putting the reason above the price

The single change that made suggestions feel trustworthy instead of just clever.

Inviting people by link instead of a username

Killed the biggest drop-off; nobody remembered handles.

A poll instead of an open chat

Groups agreed on a plan faster than they could argue about it.

Solo and safety-first modes

These came up sideways in sessions, I never ran a dedicated study on them.

Predicting traffic and booking cabs for the whole squad

A direct answer to "the 2am problem," but nobody's tried it with a real group yet.

What I chose, and what it cost

Four calls that shaped this, and what I gave up each time.

1

I built for the group, not one person.

What it cost

More work keeping everyone's phone in sync.

What it bought

That's actually how the decision gets made.

2

I explained the "why" instead of just being clever.

What it cost

Some of the slick, all-knowing feel. Early suggestions came back with just a place and a price, and felt more like magic.

What it bought

Testers didn't trust a pick they couldn't question. With one plain line (quiet till 10, live jazz after) people acted on the suggestion instead of ignoring it. Between looking smart and being trusted, I chose trusted.

3

I built two ways in, Shape tonight and Search, instead of one.

What it cost

More to maintain, and a real risk of confusing people about which one to use.

What it bought

People who already knew what they wanted stopped being forced through the mood picker.

4

When I didn't have real data, I said so instead of guessing.

What it cost

The card looks less finished. For many venues nobody has a live crowd number yet, so the card says "not available" there instead of showing a plausible guess.

What it bought

Trust. The first time someone catches the app making up a number, they stop believing any of them. I'd rather show a gap than a lie.

What's next

Where I'd take this from here.

Next · Test

Actually test the bets I'm still unsure about

SoloSafety-firstEvent diversityPredicting traffic

Right now they're just things I noticed sideways, not their own study.

Next · Edge cases

Design for the messy cases the happy path skips

A friend cancels last minuteThe venue closes earlyNobody votes in time
Next · Recruiting

Find people by how they behave, not by age or job

The best insight in this whole study came from a 42-year-old who plans dinners like a military operation, not the 22-year-old I assumed I needed.

Later · Supply

Given six more months

Let small venues, promoters and local curators list their own nights, so there's more to choose from without it getting generic. Then, quietly, something wearable.

Infinight logo

Infinight was never really about finding events.
It's about helping people feel confident enough to choose one.

The two-minute version

What I owned, what I decided, and what got messy.

Side quest · independent research and design project · Bengaluru, 2026
The ask

Self-initiated: get a group of friends from “what are we doing?” to a booked night out, and home again.

Who it's for

Groups of three to five friends in Bengaluru planning a weekend night out.

My role

Solo: research, interviews, concept, flows and the prototype.

Constraints

A solo project, so the first version had to stay small: a mood picker, a group poll and one shared tab for the night.

How I'd know it worked

How many groups reach a booked plan, time from first message to decision, and how many friends join from an invite link.

How I used AI

To prototype the app quickly, to draft and test the copy, and to critique my flows against what people told me.

Three decisions that shaped it

01

The reason sits above the price.

Options I weighed

Lead with price and rating like most apps, or with one plain line on why this place fits tonight.

Why I chose this

Testers didn't trust a pick they couldn't question. One line like “quiet till 10, live jazz after” made them act.

What it cost

Every recommendation needs a reason written for it.

02

Invite by link, not username.

Options I weighed

Find friends by username, or share one link to the plan.

Why I chose this

It removed the biggest drop-off. Nobody remembered their friends' handles.

What it cost

Less control over who joins a plan.

03

A poll instead of an open chat.

Options I weighed

Let the group discuss in a chat, or vote on three options.

Why I chose this

Groups agreed on a plan faster than they could argue about it.

What it cost

Less room for the back-and-forth some groups enjoy.

The messy part

Part of my first bet was wrong.

I designed for one person; real groups need it to work for three to five people arguing at once. And the fix wasn't a better filter screen.

What I'd do differently

  • Test with whole friend groups from the first round, not individuals.
  • Decide early how it would make money, so the design isn't built around a model that can't pay for itself.
Research, every version and the final screens

The full story, with research and every version, continues below ↓

Back to portfolio