Descent

Find the cheapest flight across every airline and airport, automatically

Project context

The problem

Finding cheap flights meant manually checking 4–6 airlines × 2–3 airport combinations × multiple travel dates — then remembering which options were cheapest across separate sessions, since prices change between visits. A single itinerary search could take 20–30 minutes of tab-switching and mental arithmetic, with no guarantee the best option had been found.

Descent eliminates the search entirely. Users set a route and a price limit, and the app checks all airlines and airport combinations multiple times a day — notifying them only when prices drop below their threshold.

How we discovered it: Personal experience as frequent travelers, validated with informal interviews with 8–10 friends and family members who travel regularly. All recognised the behaviour immediately and described the same multi-tab, multi-session pattern.

How Might We: How might we surface the most critical flight info so users can be informed about the best deals without having to search endlessly?

Goals & success metrics

Business goal: Launch with core functionality, measure organic adoption and conversion to Pro, and use early user behaviour to identify where to invest next.

User goal: User can understand a deal and decide whether to act in under 10 seconds.

What we measure:

• Price limits inserted (are users serious about setting one, or skipping it?)

• Price alerts sent out

• Price alerts tapped on

• Google Flights links clicked

• Alerts created

• Alerts created with similar/nearby airports

• Churn on recommended returns from home airport

• App Store ratings and written reviews

Design process

Early explorations

Pietro had built a functional prototype to prove the technical concept — dense, data-heavy, no visual hierarchy. I stripped it back to first principles: what does a user need to see at a glance to know whether a deal is worth acting on?

By showing testers different variations, we eliminated everything but origin, destination, alert status, and price limit + current price. All other information was moved to a detail view — specific itinerary, alternative flights and prices, and a deep-link to book. The list stayed clean; the depth was there for users who wanted it.

Key decision

During testing, the alert cards had colour-coded containers — they'd started as a Figma experiment to test colour association but made it into the beta because Pietro was moving fast. Three out of five testers said the coloured bubbles should stay, including Pietro.

I pushed back. My argument: users only preferred the colour because they'd been conditioned to it during testing — they associated colour with meaning that wasn't actually there. The cleaned design, without the distraction, would let the real hierarchy do its job.

Post-launch, the app has been consistently praised for its minimalism and legibility.

Testing & changes

We initially tested with long airport lists that included a search function. Early on I noticed that general users don't necessarily know that New York's airport is JFK, or that Istanbul has two airports — one referred to as SAW. Being a regular flyer myself, I knew these shorthand codes, but most users didn't.

We responded by adding:

• Natural language searching with on-device AI that parses airport lists and maps them to cities

• Interactive maps showing airport locations and nearby alternatives

This made the airport selection step accessible to casual travelers, not just frequent flyers.

Results

4,000 downloads within the first 3 months of launch — entirely organic, with no paid marketing and no campaign. 18 paying Pro subscribers. 5-star reviews across 20 countries.

Growth to date came from a single post by Pietro at launch and one organic mention in an Instagram roundup. Structured marketing has not yet begun.

What I'd do differently

Build a simple content and launch plan before release — even a lightweight one. The organic traction suggests real demand; a coordinated launch could have converted that into significantly higher early adoption.

/