the format is the product. three rooms — opener, main, close. always walkable. always sequenced. we get asked roughly weekly to relax this. "can't i just do one place?" "what if we want five stops?" the answer is no, and the reason isn't dogma. it's data.
we tested 1, 2, 3, 4, 5, and 7-venue plans across 200 nights between october 2025 and march 2026. each user got a randomly assigned plan length, told us a vibe, and rated the night the morning after. the curve was unmistakable.
i.one room is a meal, not a night.
one-venue plans rated highest on convenience and lowest on memory. people had a fine time. they didn't tell anyone about it the next day. one room is a transaction — you went, you ate, you left. it doesn't compress into a story you'll re-tell at the next dinner party.
this matters because the entire point of going out together is the artifact you build. nobody remembers a meal. they remember a night. a night requires at least a transition — a moment of "and then we went somewhere else." without that, the evening is a flat line.
ii.two rooms is a date, not an arc.
two-venue plans were the second-highest in convenience and surprisingly low in memory. dinner-then-drinks is a default that everyone knows. it's so familiar that the shape of it disappears. you don't remember the night because you've had the same shape of night fifty times.
the failure mode of two rooms is symmetry. you're either at place a or place b. there's no middle. there's no build. it's a binary, not a curve.
"the night needs a middle. somewhere between 'we just got here' and 'we're closing this place down.' two rooms erases the middle."
iii.three rooms is a story.
three-venue plans rated highest on memory and second-highest on satisfaction. they also produced the most "tell-a-friend" responses by a wide margin — 4.2× more than two-venue plans, 7.8× more than single-venue plans.
the reason is structural. three rooms forces an arc — opener, main, close. each one has a job:
- opener. takes the temperature down. low-stakes, easy entry, conversation starts. cold garden's patio at 6pm. a wine bar with no waitlist.
- main. earns the trust. dinner, or the kind of drinks where you're sitting somewhere nice. the room where the conversation gets real.
- close. becomes the story. the speakeasy, the rooftop, the room you wouldn't have found alone. the one that makes you say "remember when we went to —"
this isn't original to us. it's the same arc every short story has. setup, conflict, resolution. opener, main, close. the format isn't a marketing decision. it's narrative structure applied to going out.
iv.four and five are work.
at four rooms, satisfaction drops. at five, it falls off a cliff. people start texting in the middle of the night to say "we're calling it after this one." the rooms become a checklist instead of a sequence. you stop being present at room three because you're already thinking about how to get to room four.
five-room plans had the highest cancellation-mid-night rate (38%) and the lowest tell-a-friend rate of any format. seven-room plans (we ran 14 of them) all ended early.
the lesson: more is not better. a night is a finite container. cramming more rooms into it doesn't make it richer; it makes it busier. the rooms blur together.
why we won't budge.
this is the most-requested change to the product, and the one we're least likely to make. relaxing the format would let some users get exactly what they asked for and a slightly worse night. the friend who knows the city doesn't take that trade.
the format is locked. three rooms. opener, main, close. all walkable. the rooms can change. the shape doesn't.
if you're sure you want one place, you don't need us. if you're sure you want five, we're not the right tool. but if you want a night that compresses into a story you'll tell — three rooms, in the right order, by the friend who knows the city. that's what we make.