All work

2023

Flexible Flight Search for S7 Airlines

TravelTechMobileCJM & JTBD

S7 Airlines is one of Russia's largest carriers, and its mobile app is the main direct sales channel. A meaningful share of the audience was leaving for aggregators at the flight-search stage, because search only took an exact date. I designed flexible search to close that gap. The concept was designed end to end but not built.

3
Search modes designed
3 → 1
Searches to compare 3 periods
5
Journey stages mapped

UX/UI Designer on the demand analytics team. I researched the search scenarios and designed the flexible search concept end to end.

context

For S7, losing a traveller to an aggregator means losing more than revenue: it loses information about the user, weakens the S7 Priority loyalty programme, and drives up acquisition cost through partner channels.

problem

Too many iterations for a simple task

Comparing prices across three periods meant three separate searches, with no results saved between them.

Leaking to aggregators at the comparison stage

Users left for aggregators because of their flexible search and better option matching.

research

Competitive analysis

I compared S7's search against the three aggregators people were leaving for: Aviasales, Skyscanner and Google Flights.

FeatureAviasalesSkyscannerGoogle FlightsS7 (before)
Exact datesYesYesYesYes
Price calendarYesYesYesNo
Flexible dates ±N daysYesYesYesNo
Search by month or periodPartialYesPartialNo
Anywhere searchYesYesYesNo
Sort by priceYesYesYesYes

S7 was missing everything that lets you compare. It had exact dates and sorting by price, both of which assume you already know when you are flying, and none of what helps someone who doesn't: no price calendar, no flexible dates, no search by period.

Jobs to be Done

When I'm planning a holiday with flexible dates, I want to compare prices across several days or weeks at once, so I can pick the best option without repeating the search.
The primary job current S7 functionality doesn't cover, and the one responsible for the leak to aggregators. The answer: a flexible-dates mode and a trip-period mode.
When I buy a ticket, I want to be sure I haven't overpaid, so I don't end up regretting the purchase after finding something cheaper.
Even after buying, the user is left with doubt. Flexible search reduces that doubt at the selection stage: they see all the options and decide deliberately.
When I need to find the right flight, I want to do it in the official S7 app, so I don't lose the miles and status I've accumulated by going to a third-party platform.
Where user and business interests meet. The user would prefer to stay inside the S7 ecosystem, but only if the app gives them enough to work with. Right now it doesn't.

Customer Journey Map

I mapped the journey for one persona: Nikita, 28, planning a holiday in Dubai with flexible dates, who wants the best price and doesn't want to overpay. His mood runs from anticipation at the start to irritation at the point where he has to redo the whole search, and ends mixed: he buys, but isn't sure he chose well.

StageGoalPainOpportunityMetric
Opening the appGet to search and buy quicklyNo personalised destinations at the startSmart block with suggestions based on search historyActivation rate
Entering the routeEnter the route with autocomplete and suggestionsPopular destinations ignore the user's history; Dubai never appearsPersonalised suggestions by season and promotionsForm completion rate
Choosing datesFind the cheapest departure and compare optionsNo price preview on the dates themselvesPrice calendar with fares shown directly on the datesDate interaction rate
Changing the searchCheck whether Wednesday is cheaper than MondayEvery field has to be re-entered; no way to compare dates in one stepFlexible dates and trip period: every option in a single iterationRepeat search rate
PurchaseComplete the purchase of the best flightUncertainty: there may be something cheaper. Next time, an aggregatorPrice-drop alert after purchase as a trust mechanismSearch → Purchase CVR

From those pains I framed the questions the design had to answer.

  • How might we let a user compare prices across different dates on one screen without re-entering their data?
  • How might we help a user with a flexible schedule find the best combination of outbound and return dates?
  • How might we cut the number of search iterations to one without losing flexibility?
  • How might we reduce the feeling of uncertainty after purchase?

The map pointed at more than one fix. I went for the one aimed straight at the leak, comparing dates without searching again, and left the rest for later, personalised suggestions and price alerts among them: they would make the app nicer to use, but would not stop anyone leaving for an aggregator to compare prices.

design

One search, three modes

I built it as three modes of the same search rather than a separate feature. They sit behind one switch at the top of the form, so two new ways to search did not add a new screen, and someone with fixed dates can ignore the other two entirely.

Exact dates

Classic search: the user enters specific outbound and return dates and sees every available flight on those days. This covers the job of a traveller with fixed dates (a business trip, an event) and preserves the familiar flow for the current audience.

Flexible dates

The user sets a range for departure and return. Results are every combination of flights in those windows, sorted by price. This covers the primary job: comparing options across different dates on one screen without repeat searches. Sorting by price by default aligns with what the user is actually trying to do.

Trip period

The user picks a month or period plus a trip length, and the system assembles every matching outbound and return combination. This covers the job of a traveller asking when it is best and cheapest to go away for a week in July, on a holiday with a fully flexible schedule.

validation

Recommended validation

The concept stopped at design, so this is how I would have tested it.

  • A/B test: group A on current search with exact dates, group B on search with all three modes. Primary metric: Search → Purchase CVR.
  • Exit survey asking why the purchase wasn't completed, to measure the leak to aggregators.
  • Test duration: a minimum of two weeks to reach statistical significance.