I’ve spent the last few years designing inside other people’s rulebooks. Payroll under UK pay regulation, pensions under the Pensions Regulator, illegal sewage spills under Ofwat, worksite safety under OFGEM and the HSE. Different industries, different rule sets, broadly the same lesson, and it isn’t the one I expected when I started: the regulation isn’t a constraint to work around. It’s the brief.

The trap most designers fall into

Designers arriving at a regulated industry tend to walk into the same trap. They treat the regulation as scaffolding around the real design problem — an unfortunate inheritance from policy and compliance that limits what they can build. The work then becomes managing the constraint. What’s the minimum we can do to keep legal happy? Where do the disclaimers go?

This produces designs that are technically compliant and deeply boring. They tick the boxes and miss the point. Customers feel processed. Employees feel like they’re translating between two languages. Regulators see the boxes ticked but don’t see the outcomes they wrote the rules to produce.

The trap is reading the regulation as an obstacle when it’s actually a user need in disguise.

The reframe

Every regulation I’ve worked under was written because something went wrong and someone wanted to make sure it didn’t go wrong again. Strip away the legal language and the rule is almost always pointing at a user need that’s been articulated badly.

“All edits must be auditable” sounds like compliance overhead. It’s actually a finance team or production accountant saying I need to figure out who changed what when something goes wrong.

“Spills must be verified within X hours” sounds like a bureaucratic deadline. It’s actually a regulator saying we don’t want a river contaminated for two days while someone shuffles paper.

“All incidents must be reported” sounds like form-filling. It’s actually an HSE manager saying we cannot reduce harm if we cannot see it.

Read the regulation backwards — from the outcome the regulator was trying to produce, not the rule they wrote to produce it — and the design opportunity opens up. The rule is a target; the user need it points at is the brief.

This is the move that’s earned its keep across every regulated project I’ve done.

Three times this paid off

Three quick cases. Full case studies linked.

Yorkshire Water · spill verification

The regulatory rule says you must verify illegal sewage spills quickly and report them. The brief most people read is build a tool that helps analysts work through the queue faster. The brief I read — after unpacking the discovery brief with Yorkshire Water — was the bottleneck isn’t speed, it’s confidence in the signals. Each verification was taking 36 hours because analysts had no structured way to weigh on-site telemetry against rainfall data against maintenance logs.

I designed a three-tier confidence framework — absolute, high, low — that gave a stochastic ML model a structured way to weigh inputs, and gave the analysts a structured way to explain its outputs. Verification time dropped from 36 hours to about 3 minutes, model in the loop.

The pitch that won Best Pitch at the NWG Innovation Festival 2022 didn’t lead with the model. It led with regulatory risk and public harm. The room was full of water-industry people who’d been told for years that the verification problem was an analyst capacity problem. Reframing it as a signal-confidence problem was the move.

EP film & TV timesheets · payroll

UK payroll regulation requires accurate records, timely payment, and a defensible audit trail. The first version of the product was technically compliant — every edit was logged somewhere in a database — and was so unusable it was discontinued. Customers left. Retention was zero. The “audit trail” requirement had been treated as a database concern: data goes in, can be retrieved on request, done.

When I rebuilt it, I treated the audit trail as a user feature. Every edit became timestamped and attributed visually, on the approval screen. Approvers could see at a glance what had changed since last week and who had changed it. Same data, completely different role: instead of satisfying a regulator’s audit requirement, it was the thing that let production managers approve a full crew in an afternoon.

Completion time fell from 60 minutes to 5. Approval time halved. CSAT rose from 2.8 to 4.5/5. UK market share grew from 3% to 7% ahead of the TPG sale. Retention rebuilt from zero to 93%. The compliance feature was the UX feature.

Fyld · worksite safety

UK construction and utilities injuries have plateaued for eight years. In 2021 alone, 15 fieldworkers died and over 2,000 were hurt badly enough not to return to work. HSE regulation requires the reporting of incidents and near-misses, and most utilities treat that reporting as overhead — paperwork their fieldworkers grudgingly fill in. The result is staggering data loss: SGN captured an estimated 2% of the indicator events they should have seen; across the industry the annual shortfall ran to 33.6 million events. Without that data, no system can learn what actually causes harm.

Working with seven utility partners, I co-designed a standardised safety matrix with HSE managers that turned subjective recall into a structured signal. The matrix wasn’t a compliance form. It was a thinking tool — a way of categorising what had happened that fieldworkers found useful in its own right. Reporting rates moved. We tested 15 candidate ML models with two passing the 50% recall threshold, secured £460k in OFGEM Strategic Innovation Fund grants across two stages, and filed a patent application with me named as inventor.

The same instrument that satisfied the regulator’s reporting requirement was the instrument that captured the data needed to actually reduce harm. Reporting and prevention were the same feature.

The pattern in all three: the regulation wasn’t a constraint we worked around. It was the spec we worked toward.

What changes if you believe this

A few things change in practice when you take the regulation as the brief.

You spend more time with the regulator’s actual outcomes than with the rule itself. Ofwat wants clean rivers. The HSE wants nobody hurt. OFGEM wants safer streets. The rules are downstream of those outcomes; design upstream of the rules.

You spend more time with the compliance team — and discover they’re not the obstacle they’re often painted as. Compliance teams are the people who translate the rules into operating reality. They know which interpretations are required by law and which are inherited folklore. Most of the design space lives in the folklore, which means most of the design space is negotiable.

You measure the regulated outcome, not the rule. EP didn’t measure compliance — they measured timesheet completion time and customer retention. Yorkshire Water didn’t measure rule adherence — they measured spill verification time. KPIs anchor on the human outcome the regulator was after.

You stop treating compliance as a cost centre. Compliance features are the load-bearing features, in the same way that load-bearing walls are the ones that hold the building up. They’re not the wallpaper. They’re the structure.

The portable version

Regulated industries are sometimes treated as the boring cousin of consumer design — slower, less visible, less glamorous. That’s true if you read the rules as obstacles. It stops being true the moment you read them as briefs.

The work I’m most pleased to have shipped has been inside someone else’s rulebook. Five-minute timesheets, three-minute spill verifications, patent-pending safety matrices. None of those exist without the regulation that created the need they fill.

Compliance is the brief. The rest is design.