NEXT CASE
2025
Present
Head of Design @ The Times [current]
A delivery function, rebuilt into a product organisation.
The Times was running as a delivery function when I joined. This is how design, product and business analysis rebuilt it into an organisation that decides what is worth building, and then ships it well.
My role
Head of Design
Timeline
2025 → Present
Team Composition
Twelve designers across content, customer and monetisation, from three inherited
LINK
The Times and The Sunday Times are subscription news products with a 240-year-old masthead attached, which means every design decision is judged against editorial authority as well as usability. Design sits alongside product, editorial and engineering, serving millions of readers daily across web and native applications.
If you only read one part
Design here was a service function, summoned late and judged on screens.
I replaced the process with four agreements about what design and product expect from each other, then wrote the nine documents that make them enforceable.
The Times had designers. It did not have a design organisation. Everything since has been building one, with the team rather than for them.
Where I started
Design and product were not a product team. They were a delivery team, and more precisely a fire brigade: shipping as much as the capacity allowed, defensively, because the rest of the market was moving faster with more craft and more appetite.
None of this was anyone's fault, and it is a very ordinary place for a product organisation to be. It is also a ceiling.
No clear principles
No shared design system
Design summoned late, rather than embedded from discovery
Research and competitive analysis happening occasionally rather than by default
A process that constrained rather than enabled
No visibility for design above the product layer
Three designers left from the previous team
A culture comfortable with limitations, not ambition
The diagnosis underneath all of it: the company understood designers as capacity, and at best as a capability. Nobody was treating design as an organisation with a purpose of its own, which is why every conversation about it was a conversation about output and artefacts.
The reset
Made the hard call in month one.
A few contractors were out. They were not doing the job the team needed, and the longer that stayed true the more it defined what the standard was. Removing them was the first signal that the standard had changed, and it was read as one.
Built the purpose with the team, not for them.
Two months of workshops with the designers and with the product leadership team. Culture, ways of working, beliefs, manual of me, perspectives. I manage macro rather than micro, so the questions were deliberately theirs to answer.
What could we do together?
What do you want me doing for you as designers?
What kind of team do you all want to work in?
Where do you want your career to go?
Reset what the work was for.
Are we doing the best work, the best thinking, the best interface we are capable of, or are we delivering tickets? That question had not been asked out loud, and asking it in a room rather than in a document is what made the answer stick.
The agreements
A process tells people what to do. An agreement tells them what to expect from each other.
Process gets ignored the first time a deadline moves. Agreements hold, because breaking one is visible to everybody it was made with.
01.
Embedded, not summoned
Designers sit inside product teams from discovery through delivery.
The effect: design questions get asked while they are still cheap to answer, and late reversals drop.
02.
Validated, not assumed
Research, testing and competitive analysis baked into every project rather than argued for each time.
The effect: nobody has to win a debate to justify finding out.
03.
Visible, not buried
A weekly digest to the product leadership team, and a monthly review in front of senior leadership.
The effect: design stopped being a cost line and became a source of information the business relies on. It sounds administrative. It is the most consequential thing on this list.
04.
Documented, not improvised
Principles, patterns and standards owned and maintained by named people.
The effect: decisions survive the person who made them.
They are contracts in two directions. With designers they set ownership and accountability. With the business they set what design will and will not be asked to do. Autonomy without a contract is abandonment, and the playbook is what made it safe to let go.
The handbook
Four sentences carry the culture. They do not settle an argument about who signs off on a design, or what counts as finished, or what a mid-level designer has to demonstrate to become senior.
The index
Purpose and Values
Decision authority, feedback culture
Design Principles
Seven principles, with business impact
Ways of Working
The team guidance on how to work
Definition of Ready and Done
Three stages, three quality gates
Career Framework
Five levels, junior to principal
New Joiners
30, 60, 90 days, with milestones
Gradient Framework
How work gets presented to stakeholders
Prompt Playbook
Twenty prompts, and five exclusions
Design Agreements
The meta-document that holds it together
This is where expectations become contractual. A designer can read exactly what the next level requires rather than being told at review. A stakeholder can read what done means before the work starts, which removes most of the arguments that used to happen at the end.
The Gradient Framework governs how work is presented to stakeholders, so a designer is never improvising the hardest twenty minutes of the job. And the Prompt Playbook is as much about the five exclusions as the twenty prompts: the things we have agreed AI does not get to touch.
What's changed
Design stopped being capacity and started being an organisation with a purpose.
That sounds abstract until you watch it happen: designers running their own strategy sessions, bringing their own what-if work, and presenting to directors in their own right. The company did not previously believe design had a strategic position. It does now, because it has watched designers argue for one.
Design, product, editorial and engineering now move as one, with less handoff friction and faster decisions. That is an organisational outcome rather than a design one, and it came from the agreements rather than from any individual project.
We think together and we deliver together. Where the thinking is not shared, the product is just a ticket someone closed.
What I brought in
Behavioural analytics, in the room when decisions get made.
ContentSquare onboarded, so decisions are argued from behaviour rather than opinion. The change is not that we have data. It is that the data arrives early enough to change what gets built, rather than late enough to explain what already shipped.
A research repository instead of research events.
Insights synthesis and repeatable competitive intelligence in one place. Research that lives in a repo compounds. Research that lives in a deck evaporates the week after the readout.
A design system, so the team stops rebuilding the same things.
Live across web and app, with semantic naming so components carry meaning rather than labels. It earns its place here for one reason: it bought back the time the team needed for the thinking.
AI systems
AI in the workflow, with a rule about where it does not belong.
Claude, Figma MCP and Copilot in the design workflow. Prototype turnaround from days to hours, with designers shipping production-quality prototypes from VS Code rather than handing a static file to an engineer.
The fake efficiency trap.
Where it delivers: prototyping speed, maintenance, repetitive production work, documentation, exploring several directions at once.
Where it does not, yet: complex flows, novel interaction patterns, anything where the answer is not already in the training data.
The trap: AI makes it easy to produce more output, which is not the same as more value. Without rigour you get faster mediocre work. So we measure what we ship, not what we generate, and AI goes where it compounds rather than where it merely accelerates.
Agents that work inside our own rules.
Agent instructions with governance gates, so constraints are enforced by tooling rather than by me noticing. A semantic system also gives AI prompts something authoritative to reference, which keeps output on brand and reduces hallucination.
Capability over tools.
Every designer trained in AI-assisted design workflows. Tools change every quarter, capability compounds for years, and the investment is in the team's ability to evaluate, adopt and discard tools rather than in being locked to any one of them.
The outcome
from capacity
to design org
the change everything else was in service of.
Days → hours
prototype turnaround
Weekly digest
design reporting to leadership
The two weeks of questions are the part I would repeat anywhere. Every change after that was something somebody in the building had already told me, which is why almost none of it was resisted.
The agreements are the part that travels. They cost nothing, they are impossible to misread, and they changed how four functions work with each other before a single new tool arrived.
And the AI position is the part I would defend hardest. Faster output is not better work, and a design function that cannot tell the difference will spend the next two years generating things nobody needed.