CommBank
20_
Product Design
_24
FuelFinder ↓
In 2024, Australians were paying more for fuel than they needed to. Research showed that CommBank customers were turning to fuel priceing apps to find cheaper prices, but most weren't, and those who did had to do all the work themselves. We saw an opportunity to change that.
My Role ↓
In 2024, fuel costs were one of the most visible pressures of rising cost of living in Australia. NSW customers were losing around $20 per tank to price differences of up to 40¢/L between stations less than 3km apart, not because cheap fuel wasn't available, but because finding it in the moment was too much effort.
Existing fuel apps put the work on the customer. They required people to take the initiative, open an app, search for the best price, and act on it themselves. Most people didn't bother.
CommBank had something those apps didn't, a view of customer spending patterns that could surface the right information at the right moment, without customers having to go looking for it.
By using CommBank's AI-powered transaction analysis to predict when a customer would next need to refuel, we could notify them at exactly the right time with the cheapest nearby option already identified. No searching, no scrolling, just a timely nudge that did the work for them.
Rather than design a full product and ship it, we deliberately started small. The hypothesis was simple: if we can show customers a relevant, timely fuel alert, will they act on it? We ran three progressive experiments to find out, each one building on what the last one taught us.
Experiment one
Does the idea work at all? We stripped the concept back to its simplest form — a single push notification telling customers where the cheapest nearby fuel was, right now. If they tapped it, it opened directions to the station. Nothing more.
The experiment ran for five weeks across 50,000 customers. The point wasn't to build a product. It was to find out if a well-designed, well-timed nudge was enough to change behaviour. Customers engaged. That was enough to keep going.
Experiment two
Does a fuller experience hold up? With desirability validated, we designed and built a production pilot inside the CommBank app with predictive alerts, a fuel station map, and price comparisons. We released it to 250,000 pre-selected customers over six months.
This was the first time we could see how customers actually moved through the experience end to end, and it gave us a clear picture of where it worked and where it didn't.
Experiment three
Can we build something that lasts? We analysed everything the pilot taught us, usability issues, navigation patterns, and where people dropped off. The information architecture was rebuilt to support new features, and the visual design was aligned with CommBank's updated design system. We then released to all NSW customers.
Reflections
I was fortunate to own the project end to end, conducting customer research, collaborating with product managers to define product strategy, facilitating ideation, designing and prototyping, and working closely with engineers to deliver a production-ready experience.
Running experiments let us build confidence progressively, only investing more when the previous step proved its worth. By the time we built the full product, we already knew customers wanted it. For a zero-to-one feature inside a major bank, that matters. The compliance overhead, engineering time, and stakeholder buy-in required to ship something new is significant.
Over ten months from June 2023 to March 2024, Fuel Finder saved CommBank customers almost $2 million. More than 109,000 people opted in, generating 664,000 fuel transactions through the product and saving an average of 5.45% per fill — roughly $199,000 in collective savings every month.
