1,380 NYC listings · 24 neighborhoods
1.41% vacancy, lowest since 1968 (NYC HPD)
Your match % is a metric
01The problem
A bad roommate is a daily tax on your sleep, your kitchen and your peace.
Most roommate searches end with one DM vibe check and a 12-month lease. The mismatch shows up after move-in.
Sleep
Their 6 a.m. alarm, your 2 a.m. bedtime.
RoomMe asks
“What time do you actually go to bed on weeknights?”
Dishes
The sink that’s never empty.
RoomMe asks
“When do your dishes usually get washed?”
Guests
The partner who quietly moved in.
RoomMe asks
“How often does someone stay the night?”
Noise
Calls on speaker at midnight.
RoomMe asks
“How loud is your home on a weeknight?”
02Neighborhoods
Start with where you want to live.
See all 24 neighborhoodsPublic listings as of Sep 26, 2026. We link back to the source.
03Why you can trust it
Roommate matching that can explain itself.
Metric
Your match % is the average agreement across your 10 answers. AI never sets it.
01 / 05
87%Match
How it’s scoredBedtime100%
You 23:00–00:00 · Sam 23:00–00:00
Noise75%
You quiet talking · Sam music or TV, low
Overnight guests64%
You 1–2 nights a month · Sam about once a week
Kien and Sam are sample profiles.
04How it works
From pre-screen to House Agreement
1. 30-second pre-screen
Budget, move-in, neighborhoods, dealbreakers.
2. Tap, then talk
10 habit questions, then talk for 90 seconds. Typing works too.
3. Shortlist up to 5
Every match % explained. A meetup opens only on a mutual yes.
4. House Agreement
Quiet hours, guests, chores and bills, agreed before the lease.
The math
How your match % is calculated
Three steps, the same for everyone. The table is read straight from the scoring code.
- 01Dealbreakers filter firstChecked both ways, before any scoring. “No smoking/vaping indoors” rules out anyone who smokes inside. A pet allergy, or needing a pet-free home, rules out anyone with a pet or planning one.
- 02Each answer becomes a real quantityBedtimes become clock hours, guests become guests a month, noise becomes decibels. Then, per question:agreement = max(0, 1 − gap ÷ full-clash gap)The same answer scores 100%; the full-clash gap or wider scores 0%. Rounded to a whole %.
- 03Your match % is the averageThe mean of the 10 agreements, rounded. Every question counts the same: 10 points. AI never sets it, and voice answers don’t change it.
The 10 questions, in real units
| Question | Value per answer | Gap | Full-clash gap |
|---|---|---|---|
| Bedtime clock time | 21:30 Before 22:00 22:30 22:00–23:00 23:30 23:00–00:00 00:30 00:00–01:00 01:30 After 01:00 | Hours apart on a 24-hour clock | 4 h apart |
| Wake time clock time | 05:30 Before 06:00 06:30 06:00–07:00 07:30 07:00–08:00 08:30 08:00–09:00 09:30 After 09:00 | Hours apart on a 24-hour clock | 4 h apart |
| Cleaning times a week | 0 0 1 1 2 2 3.5 3–4 6 5 or more | Ratio of (1 + value) | 4× apart |
| Dishes hours until washed | 0.5 Right after eating 8 The same day 24 The next day 72 When the sink is full | Ratio of (1 + value) | 64× apart |
| Guests guests a month | 0 0 1.5 1–2 4 3–5 8 6–10 14 More than 10 | Ratio of (1 + value) | 8× apart |
| Overnight guests nights a month | 0 Never 1.5 1–2 nights a month 4.3 About once a week 10 2 or more nights a week | Ratio of (1 + value) | 8× apart |
| Sleep noise reports it or not | Yes I snore Yes I grind my teeth Yes I talk or move in my sleep No No one’s mentioned it Yes Other | Same answer or not | None: a mismatch scores 50% |
| Noise dB | 30 Silent 45 Quiet talking 55 Music or TV, low 70 Music or TV, out loud | dB apart (approx.) | 40 dB apart |
| Work from home days | 0 0 1.5 1–2 3.5 3–4 5 5 or more | Days apart | 5 days apart |
| Smoking steps | 0 Never 1 Outside only 3 Sometimes inside 4 Often inside | Steps apart | 4 steps apart |
Counts use ratios
Never to once a month is a bigger change than 8 to 9 times a month.
Sleep clashes at 4 hours
4 hours apart puts one person’s wind-down or morning inside the other’s sleep.
Smoking skips a step
“Sometimes inside” sits at 3, not 2: crossing indoors counts double.
Sleep noise costs half
Snoring or grinding can be medical, so a mismatch scores 50% and never rules anyone out.
Worked example
You and Sam
“You” is Kien. Both are sample profiles, scored by the app’s own code.
87%
match · metric
Bedtime100%
You 23:00–00:00 · Sam 23:00–00:00
Wake time75%
You 07:00–08:00 · Sam 06:00–07:00
Cleaning100%
You 2/week · Sam 2/week
Dishes100%
You the same day · Sam the same day
Guests100%
You 1–2/month · Sam 1–2/month
Overnight guests64%
You 1–2 nights a month · Sam about once a week
Sleep noise100%
You no one’s mentioned it · Sam no one’s mentioned it
Noise75%
You quiet talking · Sam music or TV, low
Work from home60%
You 1–2 days · Sam 3–4 days
Smoking100%
You never · Sam never
Match %(100 + 75 + 100 + 100 + 100 + 64 + 100 + 75 + 60 + 100) ÷ 10 = 87.4 → 87%
FAQ
Questions worth asking first
No. It’s plain math on your 10 quick-tap answers, the same with AI on or off. Voice answers never change it.
