September 29, 2026 · 4 mins read

What 44 Years of Service Design Can Teach Us About the Next 10

Ben Rose_.SDinGov

Making a service map has never been easier. With AI, you can have a blueprint, a prototype and a slick journey diagram before lunch. But the making was never really the hard part. The harder question is whether any of it ever reaches the people who need it.

Ben Rose, Client Partner at GAIN, explored this question in his talk ‘Evolution of a craft: a short history of service design’ at SDinGov in Edinburgh. Ben spent around ten years leading service design at Methods, working on roughly 100 public sector projects, and he squeezed the discipline's 44-year story into 45 minutes.

It all goes back to 1982, when Lynn Shostack published the first service blueprint in the European Journal of Marketing. She's often called the mother of service design, though the field has plenty of other parents. Marketing handed us blueprints. Theatre gave us front stage and backstage. Anthropology brought fieldwork, and systems thinking brought ecosystem and galaxy models. Service design has always been a magpie, and it's better for it.

Ben walked through how the tools have changed, using his own projects as the thread. In Coventry in 2017, blueprints were flat files sitting on one person's laptop. A year later at the Department for International Trade, they'd grown into whole ecosystems of blueprints built around jobs-to-be-done. Then in 2019, working with DHSC on the Medical Examiner Service, the team put a physical blueprint in the room and built it alongside the policy team from the very first day. The work went on to shape legislation. On 9 September 2024, the service became statutory across England and Wales, 21 years after the Shipman inquiry first called for it.

Sometimes, though, the trusty blueprint just doesn't fit. NHS cardiology in 2022 was one of those times. There was no obvious start or end to the problem, so Ben's team built a galaxy model instead, with five solar systems: emergencies, assessment, treatment, knowledge and logistics. The planets were the fixed, action-led parts of care, like recognising symptoms. The moons were the current ways those parts get delivered, and those could move. Rather than redesigning cardiology from scratch, the advice was to make each part better and improve how they connect.

Zoom out and there's a bigger shift going on. Early digital services were built around transactions, but nobody gets out of bed wanting to complete one. Lou Downe summed it up in a 2015 blog post: "Good services are verbs, bad services are nouns." The Service Standard went on a similar journey, slimming down from 23 points in 2013 to 14 in 2019 and quietly losing the word "digital" along the way. The big change came at point 2, which now asks teams to solve a whole problem for users, even if it means working across organisational lines.

None of this is especially new. Tell Us Once, the service letting you report a death to several parts of government in one go, was first recommended in 2006 and went national in 2011, before GDS even existed. And there's still a long road ahead. A GDS review in January 2025 found 47% of government services have no digital route at all.

Then came AI. Ben told us about an experiment where he trained Claude on ten years of his own service design work and asked it to create a tube map for a made-up skateboarding licence service. The result looked great. What it couldn't do was choose a direction, spot a road no longer there or read the room. In Ben's words, "AI democratises execution; it doesn't democratise judgement. Yet."

Which is why storytelling matters so much. At Methods, Ben made it the number one thing he hired for. A map pitched to a product owner has to land very differently when it's in front of a deputy director, and no tool is going to make those calls for you.

Ben finished with the story of Elvis and Jerry. Jerry fixed Elvis's problem: from the stage, he couldn't see a single empty seat. But 5,000 people still weren't in the room. As service designers, we've got good at changing what departments can see from the stage. We've been far less good at getting our maps into the hands of the person who's actually lost.

In public services, there's no competitor down the road and no second night. AI will keep changing how we work, but it won't change why the work matters. The only real move is to make the map, then hand it to the person who needs it.


Recent Posts

See All