Ryanair · London Stansted
The bags that forecast fuel
A boarding-delay problem turned out to be a fuel-forecasting problem.
- SymptomBoarding delays at London Stansted
- Root cause189 seats, overhead room for about 100 trolley bags
- First fixFrom bag 101, cabin bags go to the hold at the gate
- ChangeThe 10 kg cabin bag becomes a paid product
- ConnectionThe airline now knows who carries one: 85 kg or 75 kg per passenger
- PayoffSharper fuel forecasts and fuel bought ahead at a known price
Problem
Ground crews at Stansted were seeing serious boarding delays. I was the product manager of the team in charge of finding root causes and fixing them.
What we saw
Most passengers flew with a trolley bag, and the bins only fit about one for every two seats. At the same time, a Business Intelligence team next to us was working on fuel savings with a model that gave every passenger the same weight.
What we did
First, bags beyond the hundredth went to the hold at the gate, at no cost. When the 10 kg bag became a paid product, the weight model could change too: 85 kg for passengers who paid for the bag, 75 kg for those who didn't.
Result
Smoother boarding and fewer delays. Fuel estimates precise enough to plan years ahead and buy fuel at a known price. And almost pure-margin revenue from a bag that used to fly free.
TakeawayTwo teams were working on the same problem without knowing it.
Ryanair · Ryanair.com
Rebuilding the shop without closing it
We moved the booking funnel of a site serving 149 million customers a year from a monolith to microservices, one page at a time.
- Page 1Home
- Page 2Flights
- Page 3Bags
- Page 4Seats
- Page 5Extras
- Page 6Car hire
- Page 7Payment
Problem
The site ran as one large monolith, which made it hard to scale and slow to take into new markets.
What we did
We split it into microservices in phases, following the sales funnel: home, flight selection, bags, seats, extras (airport transfers, inflight, insurance, Fast Track), car hire and payment.
Result
A site that scaled, dozens of new markets opened almost at once, and a booking flow that became a pattern across European low-cost carriers.
TakeawayYou can rebuild a product with enormous traffic without breaking it.
ING · Credit card
The free card nobody used
Customers rated our credit card highly. They just didn't use it.
- SymptomA free card customers rated well, but rarely used
- DiscoveryFree instalments on big purchases could make it an everyday card
- BlockerThe instalments engine had to be rebuilt first
- LaunchBuy Now Pay Later, end to end
- Discovery againCustomers asked for flight points and cashback
- IterationPoints were blocked by airline partnerships, so we signed cashback partners
Problem
The card was completely free and compared well with the rest of the market. Customers needed a reason to make it their everyday card.
What we did
Discovery pointed to free instalments on large purchases. Technical limits forced us to rebuild the instalments architecture before launching Buy Now Pay Later, which cost us time to market. After launch we went back to customers, and their answer led to cashback partnerships.
Result
- +17% card volume year on year, two years running
- +15% card transactions
- +13% new-card activations
- In parallel, −20% operational incidents through root-cause work
TakeawayDiscovery isn't a phase. We ran it again as soon as the first answer shipped.