Case study 1 of 4, travel app
All the research behind a good trip, so you can just pack and go
Raasta helps Indian travellers find offbeat places and gives them what they need to actually go: when, how, what it costs, and what could go wrong. I founded it and I’m the only designer on it. This is how it went from a big, messy idea to an app I could hand over to a developer.
- Role
- Founder and product designer
- Team
- Me on research and design, Siddharth on dev
- Platform
- iOS and Android
- Scope
- Product, UX, UI, design system, brand
- Timeline
- 4 weeks, finished Nov 2026
01
Offbeat places are hard to plan for, so people don’t go.
Say you want to go to Chitkul in Kinnaur. The useful bits are spread across four places, and you can’t tell what’s still true. Most people with a few days off and a fixed budget don’t bother. They go to Manali again.
Apps like Tripoto and Holidify are built around articles. You read a long post and pull out the parts you need. I wanted Raasta to feel more like Zomato or Airbnb: you scan, compare, save, and come back.
The goal: open one place and know whether it’s for you, when to go, how to get there and what it’ll cost, without leaving the app. Raasta stops short of booking.
02
Where people plan today, and what each app gets wrong for this.
Where the information lives
Reddit threads, YouTube vlogs, Instagram reels and old blogs. Each one has a piece. None of them tells you if it’s still true. That made trust the core problem, more than discovery.
Who it’s for
Indian travellers around 22 to 35, mostly from Tier 1 and Tier 2 cities. Short trips, tight budgets, often with friends. They plan on their phones, sometimes on patchy networks, and they’re sceptical of polished travel content.
Why not Instagram?
It’s great at making you want to go somewhere. It’s bad at being the place you come back to when you actually plan. Raasta is built for the second part.
| I took apart six apps | What I took | What I did differently |
|---|---|---|
| Airbnb | Browse without an account, log in only to save | No booking, costs as ranges |
| AllTrails | Scannable difficulty, trust from reviews | No “visited on” dates |
| Komoot | Highlights as the hook | Short “why people go” lines |
| Zomato | Fast scanning, location context | Location only changes travel costs, never what you can browse |
| iOS Weather | A clean 7-day strip | Only stats that matter for a trip |
| Threaded replies | No karma, softer threads |
Rules I had to design around
Before the first screen, I listed what the App Store, Play Store and India’s data protection law need from an app like this. These shaped onboarding and Settings from day one.
- Account deletion inside the app
- Sign in with Apple next to Google
- A consent notice
- A way to report comments
03
The first version promised too much.
My first version of Raasta had a score for every place, two kinds of multi-day trips, hour-by-hour plans, shareable collections and a follower system. It looked great on paper. It was also more than two people could build or keep accurate.
So before designing screens, I went through every feature and asked one question: can we keep this correct for 150 places, with no research budget and no team on the ground?
- A score for every placeA 12-month view and three quick factsOne number hid too much. Chitkul is great in May and closed in January.
- Day-by-day plansGone for single placesPeople plan their own days. Keeping plans right for every place would take more effort than anything else in the app.
- Multi-stop journeysSingle destinations onlyMulti-stop trips come later, built by our team, not by users.
- Collections and share linksPersonal savesNice to have, not why anyone opens the app on day one.
- Followers and followingA profile about trustA social graph needs people first. The profile shows your comments, helpful votes and where you’ve been.
What’s left is small: a feed to browse, a detailed page for each place, saves, and comments where travellers can correct us.
04
I designed the destination page first, and built everything else from it.
It’s where Raasta earns trust or loses it. The feed card, the saved list and search results are all smaller versions of it, so it made sense to get this one right and then shrink it.
The top answers one question: is this my kind of place? Photos, name, state, mood tags, a short description and three quick facts. Below that, four tabs, each for a different question.
- 01OverviewShould I go, and when?
- 02ExploreWhat will I do there?
- 03PlanHow do I get there, and what should I know first?
- 04CommunityWhat do people who’ve been there say?
The photo stays where it is and the page slides up over it. When the content reaches the top, the floating buttons settle into a solid bar with the place name, back, save and share, and the tabs pin right under it. It happens as one continuous move, with no jump or reload. From there, every tab is one tap away and you can always go back or save, however far down you are. It works the same on all four tabs. Switching keeps the bar and tabs where they are and only changes what’s under them. The pinned part takes 168 of the screen’s 874 points, so the content still gets over 80% of it.
I tried about nine tab styles and picked tabs with counts, so you know Explore has 12 spots and Community has 24 comments before you tap.
05
Seven decisions that shaped the app.
01 When to go
Quiet months shouldn’t look dangerous.
The obvious answer was a green-to-red scale. When I tried it, months that were just off-season looked like warnings. That’s not true, and it scares people off places that are perfectly fine to visit.
I settled on 12 dots and four states: Best, Good, Avoid and Closed. Closed is a dashed outline so it reads differently from Avoid. The current month gets a ring. Seasonal warnings like monsoon landslides moved to their own list.
When to go
Best in May–Jun and Sep
- Best
- Good
- Avoid
- Closed
What it cost: Less nuance. Four states can’t show that late May is better than early May.
02 Trip cost
One cost number is wrong for most people reading it.
A trip to Chitkul from Delhi costs a lot less than from Lucknow. So travel is worked out from your home city, while stay and food are the same for everyone.
This took more rounds than anything else in the app, over eight. I also tried a Budget/Comfort switch and a Solo/Duo switch, and removed both. Being wrong about money is the fastest way to lose someone’s trust.
Trip cost
₹6,000 – 12,000
Per person, about 4 days · ≈ ₹1,500–3,000 a day
- Getting there₹3,500 – 6,000Bus or train, both ways. Changes with your home city.
- Stay₹1,500 – 3,000Homestays and guesthouses. Same for everyone.
- Food and local₹1,000 – 3,000Meals, local rides, activities. Same for everyone.
Rough estimates for one person, based on what travellers have shared.

What it cost: No exact prices. Ranges feel less impressive than a precise number, but they’re honest.
03 Trust
Show what’s checked. Don’t label what isn’t.
Every place is checked by our team before it goes live, so it gets a small “Verified · Oct 2026” badge. A “312 visited” pill only shows after 25 visits, so a new page never shows an awkward “2 visited”.
- No visit dates on comments. We can’t verify them, and asking feels intrusive.
- No “Corrected” badges. When a traveller points out something wrong, we fix the guide quietly.
- No nearby hospitals. If that’s wrong, someone gets hurt. Altitude advice is marked as general information instead.


What it cost: Fewer “proof” signals on the page. I’d rather show less and have all of it be true.
04 Community
The comments are how the guide stays up to date.
We can’t visit every place every season. People who’ve just been can. Comments are how we find out a homestay closed or a road reopened.
A classic thread beat a highlights carousel and a topic list, then went through about six rounds. Pills, chips, dividers and thread lines were all fighting the text, so I stripped most of that out. The composer’s resting, typing and replying states share one shape, so nothing jumps when you tap.


What it cost: A cold start. Small places, the ones Raasta is really for, might sit empty for a while.
05 Accounts
Browse freely. Log in only when you want to keep something.
You can skip login on first launch and browse everything. After that, it only comes up when you save a place or post a comment, and you land back where you were, with the place already saved.
I dropped a “what’s your travel vibe” quiz from onboarding. Your mood changes from trip to trip, so mood, season, budget and trip length are filters you set each time.
What it cost: We know less about new users on day one. Beyond a name and username, home city is the only thing we ask for, because it decides which travel costs they see.
06 Home city
Everything starts from your city.
Onboarding asks one question: where you travel from. After that, the app measures everything from there. Weekend trips are listed by hours on the road or train, budget lists are priced from your city, and filters use travel time from home. It’s the same reason the trip cost is split. Only “getting there” depends on where you live, so that’s the only part that changes.


What it cost: One more onboarding step. If you skip it or turn location off, you pick a city from a list.
07 Profile
The profile counts help, not followers.
A travel profile can easily turn into a follower count. Raasta’s leads with how many travellers found your comments helpful, then the states where you’ve helped. Under that is a grid of all 36 states and UTs that fills in as you travel. A new profile starts with an empty grid and one line telling you how to fill it.


What it cost: No follower numbers to show off, and some people want those.
06
Four tabs, each doing one job.

Home
A feed for when you don’t know where to go yet. Categories are the type of place. Moods are filters, so “Spiritual” never shows up as both.

Explore
Ways in by mood, trip length, region, or weekend trips from your city. Only tiles we can fill at launch.

Saved
One flat list, plus the places you looked at recently.

Profile
Who you are and where you’ve been. All 36 states and union territories as tiles that fill in as you travel.
Every screen on one canvas.
151 screens and states across nine flows, including errors, empty states and what happens offline. Onboarding and Profile alone account for 83 of them.
07
A small system, so new screens are easy to add.
Colour has jobs. Dark carries every button. Orange only shows state: selected, saved, active, because white on orange fails contrast. Green means good or verified, nothing else.
Tokens in two layers. 48 primitive colours under 109 semantic tokens named for what they do, not the hue. Spacing and radius have 20 more.
One typeface. 37 text styles on Google Sans Flex. A final audit found no hard-coded values outside illustrations and logos.
Components. 79 component sets with 440 variants, grouped the way a developer would look for them, all in one file.
Built for phones. No hover on a phone, so components have Default and Pressed. Sheets are for content. Anything you confirm uses the native iOS dialog.
Icons. 38 line icons from one library, plus 92 custom icon components for packing, weather, costs and categories, all on whole pixels.
08
Most of the work was in states you hope people never see.
- Delete account inside Settings. Both app stores require it.
- Report and moderation for comments, since anyone can post.
- Offline, update required, maintenance and “something went wrong” screens.
- Denied location, wrong codes, taken usernames and the rest of what goes wrong at login.

09
One designer, a lot of options, not a lot of time.
I used Claude throughout as a research and iteration partner. It helped me look into store and privacy rules, compare how reference apps handle a pattern, pressure-test ideas before I committed to them, and audit the Figma file for drift. That meant I could explore more directions per section and still finish all 151 screens.
The direction, the trade-offs and every screen in this case study are mine. The tool made the process faster. It didn’t make the calls.
10
What I’d do differently, and what’s next.
Not tested with real users yet. Every decision here comes from comparing options side by side, not watching people use it. The waitlist is the first group I’ll test with once there’s a build.
I’d cut scope earlier. Asking “can we keep this accurate?” on day one would have saved weeks.
The cold start is real. Comments correct the guide, but only where people are commenting. Small places need another way to stay fresh, and I don’t have a full answer yet.
Next up: multi-stop trips built by our team, a shared wishlist for friends, and Convoy Mode for group trips with live location.


