A discovery and booking app built around the two moments every other travel product skips: choosing together, and the days after you’ve paid.
Travel apps are good at showing you places. They’re terrible at helping you decide.
Finding a nice hotel takes four minutes. Getting three people to agree takes four days. And the moment money changes hands most travel apps go quiet, right when you need them.
A diary study: 14 travellers, two-week diaries, 9 follow-up interviews, groups of 2–5.
Four findings that shaped everything.
Group trips stalled on agreement, not money.
Everyone moved to notes, screenshots and the group chat.
Asked privately, because nobody could see prior answers.
Saving was a reflex, not an intention.
Three decisions I got wrong first.
The reason lines made a browsing surface feel like an argument, and promised a ranking the feed didn’t deliver.
Chat collapsed three unrelated conversations into one. Split apart, each could be shaped properly.
Shadows put a layer of interface between the viewer and the place.
Five features, in the order a trip happens.
Full-bleed photography, one category strip, no interstitial noise. The only data besides name and price is a deal pill, and only when the price is genuinely below its usual rate.
Every price carries context: how it compares to the usual rate, whether it dropped since you saved it, and which nearby window is cheapest.
The shortlist argument already happens in a group chat, so the vote belongs there. A poll is a structured message with a close time and a tie rule.
The itinerary becomes the home screen. Walk times, a NOW marker, weather, offline save, and a wallet holding passes and confirmations.
No travel app gives you a line to the person who owns the room. Host questions are answered publicly, so the same question is asked once.
One accent, used only for the thing you should touch next.
faster group decision in moderated tests
called price context the reason they’d trust it
closed the app after booking, against 14 of 14 before
features cut after testing
Blind voting tested well with three people and badly with five. And host Q&A assumes hosts answer, with a slow host it reads as abandonment.
Everything in the folder.
wander-01-feed.jpg
wander-02-search-filters.jpg
wander-03-detail.jpg
wander-04-map.jpg
wander-05-discover-deck.jpg
wander-06-saved-board.jpg
wander-07-trip-planner.jpg
wander-08-document-wallet.jpg
wander-09-booking.jpg
wander-10-inbox.jpg
wander-11-host-qa.jpg
wander-12-party-vote.jpg
wander-13-poll-composer.jpg
wander-14-wander-actions.jpg
wander-15-onboarding.jpg
wander-16-disruption.jpg
wander-17-recap.jpg
wander-18-profile.jpg
wander-light-01-feed.jpg
wander-light-03-detail.jpg
wander-light-07-trip.jpg
wander-light-12-party-vote.jpg