British Gas Home Insurance
Lead UX Designer, contract. Six months from launch to every two-year KPI.
Client
British GasRole
Lead UX Designer (Contract)Timeline
2018 - 2019Outcome
25% conversion vs 12% avgTL;DR
I was brought in to build a connected-home product, ended up building Home Insurance instead, and it became the most successful venture Centrica had launched. We hit every two-year KPI in six months and held a 25% conversion rate against the 12% that price-comparison sites average. The lesson that got us there was counterintuitive: people didn't want the quick, cheap quote. They wanted the long one. And the research habits I built proving things like that are the same ones I carried into Uptime years later.
What I was actually hired to do
The original brief was Home IQ, a blend of insurance and preventive home services. I ran the discovery with the team, it tested well, and then it got handed to the Hive team who already owned connected devices. Fair call, wrong home for me. Oliver, the other UX designer, moved on to another team, and I was handed Home Insurance to build. The real ambition was bigger than one product. British Gas sold energy. The bet was that the brand's trust could carry it into services people don't associate it with, and that meant taking Home Insurance off the old third-party model and owning the experience end to end. I was the only designer on it: one PO, one BA, a scrum master, a couple of devs and testers, and me running discovery, research, UX and UI.
The assumption I got wrong
I knew how everyone else in this space operated, and I knew the British Gas brand carried trust, so my first instinct was the obvious one: make the quote journey fast. Fewer questions, quicker price, less friction. Standard playbook. It tested worse than expected. Badly, even. What I found watching participants was the opposite of the playbook. Even when the quick quote came back cheaper, people preferred the long one. The extra questions made them feel the quote actually reflected their home, their situation, their risk, not a generic number. Speed read as carelessness. Length read as care. So I did something that looks wrong on paper. We added questions the back-end didn't even need, purely so the journey felt thorough enough to be trusted.
The friction wasn't the enemy. The right friction was the product.
How I actually knew
None of that was a hunch I got lucky on. This is the project where I built the way I run research, and it's the part of British Gas I carried furthest. I treated every design decision as a hypothesis with a way to prove it wrong. Quick quote versus long quote wasn't a debate, it was a test, run with real participants, with the cheaper option deliberately on the table so a preference for length couldn't be explained away by price. I ran sessions constantly, watched live usage, and triangulated what people said against what they did and what the analytics showed, because those three rarely agree and the gap between them is usually where the insight is hiding.
The bigger thing was making research something the whole team could do, not a bottleneck that lived with me. I built a repeatable way to frame a question, get in front of users quickly, and come back with an answer everyone could act on. That process is the one I later brought to Uptime, where several people picked it up and ran their own research the same way. British Gas is where it was forged.
The 2% nobody could see in the data
The clearest proof of why I watch real usage came late. Something that never showed up in a single user test and never surfaced in the analytics: live, real users were trying to untick items on the assumptions list, the "we've assumed these things about your home" step. They wanted to correct it, and the design wasn't letting them. We let them. That one change added another 2% to conversion.
The behaviour only appeared when the stakes were real and nobody was performing for a facilitator. That's the whole argument for watching real people, in one small moment.
Where it landed
This was a big one, for British Gas and for me. Home Insurance became the most successful venture in Centrica's history, hitting every one of its two-year KPIs in six months and converting at roughly three times the old third-party model it replaced. I ran it solo through the month before launch and its first stretch live, using Adobe Analytics to track conversion and friction and live sessions to catch what the data couldn't.
A 25% conversion rate against the 12% that price-comparison sites average, built by trusting what people actually do over what the playbook says they should.