Home› The operating model› A function built from zero
Rhythms, intake, capacity, rituals, tooling and visibility are the goods. Designers, researchers, content designers, leadership and the other Ops functions are the buyers. Adoption is the transaction. And like any marketplace, the failure mode isn't bad inventory. It's a buyer who couldn't find what they needed, didn't trust it would hold, or found it cheaper to work around you.
Built this before:
The honest scope: a start-up, not a 130-person design org — and his reports were makers rather than ops practitioners. The step-up, addressed →
Every good a central function offers has a buyer, a value proposition, and a specific reason someone decides it's cheaper to do it themselves. Naming that reason honestly is most of the job. Here's how I'd think about the inventory for a 130+ person Design and Research org.
The quarterly and monthly cadence that keeps 130+ people pointed the same way without anyone attending nine meetings to find out where they're pointed.
One front door, and written principles for how the queue is ordered — so priority is a lookup instead of a negotiation.
An honest read on what the org can absorb, so commitments are arithmetic rather than optimism, and so "we're at capacity" is a number instead of a mood.
Standups, crits, retros, reviews. The connective tissue of craft — and the single easiest thing in any org to quietly ruin by adding to it.
Figma, Jira, Notion, Cursor, Claude, Google Workspace — plus whatever accumulated quietly. Chosen deliberately, adopted properly, and paid for on purpose.
Newsletters, dashboards, executive summaries. The thing that stops a central function from having to argue for its own existence every planning cycle.
The first ninety days are laid out step by step on the about page — people first, then the queue, then the calendar, then one small AI pilot. This is the work that comes after, once there's enough trust to change how the function operates.
Nothing enters prioritization until it's actionable. A backlog of vague intentions turns prioritization into a debate about interpretation instead of value. Granular enough to finish, in one place, with an owner.
Done it: at Dow Jones the definition of ready is enforced in Relay — planning, structuring, prioritization and resourcing all happen there, and its automated follow-ups chase the half-baked briefs until they're ready. The structured output then transfers through Asana's API into the execution and reporting layer I own, across ~8 events and 5 launches over the past five months. The full AI point of view →Consistency where it saves people effort, difference where the disciplines genuinely differ — and a bias toward adopting a neighbour's working practice over inventing a parallel one. Being a generous citizen of an Ops community mostly means giving away things you built.
Doing it: I'm a Brand-side program manager partnering daily with product design teams and fielding work from six functions — the seam is my current job.External work has to arrive at the standard and in a format the workflow can accept — which means front-loading the spec and testing far earlier than anyone wants to, not reviewing harder at the end.
Done it: owned the Monaverse relationship through implementation and platform testing to launch, plus external motion, 3D and video vendors — and delivered from the agency side at Moving Brands and Prophet.Every program gets an owner who isn't me. Coaching is mostly handing over something slightly too big and staying close enough to catch it. The measure of an ops leader is how much of the function runs while they're on holiday.
Done it: sourced, hired and trained the Everyrealm creative team, then handed them the process rather than the instructions.A design org's tool stack is a set of standing commitments that renew whether or not anyone is looking. Here's how I'd approach it — and where I'd want a good partner, because this is the part of the role I've done least formally.
The first artifact I'd build is unglamorous: every tool, its cost, its renewal date, its owner, and what fraction of the licenced seats are actually used. Most stacks have at least one line item that is entirely a habit.
It usually happens because a team needed something on a Tuesday. That's a legitimate reason — so the fix is to pick one deliberately and migrate the loser properly, not to send a note telling people to stop.
Procurement, Finance and Legal move fastest when you arrive with the alternatives priced, the adoption plan attached and the risk already named. I've been on the agency side of these contracts, which is a useful education in how they're really scoped.
The most expensive tool in any org is the one that was bought, announced, and never taught. Half the team keeps using the old thing and you're now paying for both.
For twenty years the deal was that you found the closest tool and reshaped your process to fit it. In the age of AI, software bends to the real business need instead. So for any workflow the market doesn't serve, "build it" is now a legitimate answer — and sometimes a days-long one, against a five- or six-figure annual line item that still wouldn't fit how the team works.
AI collapsed the cost of implementation, not the cost of thinking. So I build the traditional way: research the problem, define it, write requirements, then build. That's the guardrail that stops "we could just build it" from becoming a shadow stack nobody maintains. It also matters for stewardship: because I build with Claude and Cursor rather than evaluating them from a distance, I can tell the difference between a tool a design team will absorb and one that demos well.
Vendor management ✓. Commercial exposure from the agency side ✓. Delivery inside legal constraints during the Endo–Mallinckrodt merger ✓. Profitability via resource utilization and capacity management is a stated strength ✓. But a full Procurement and Finance cycle for a design org's stack — not yet. I'd want a strong partner through the first renewal season and I'd say so on day one rather than discover it in month four.
The complete map. Where the match is partial, it says partial — that's the point of an item-details panel.
warren·shop · Star Seller
Relay, the AI intake and planning layer he built — it turns 6 teams' inconsistent quarterly spreadsheets into one resourced plan — plus the Asana execution and capacity model behind ~8 events and 5 launches over the past five months.
Influence
Quick look — item details
warren·shop · Star Seller
4–6 designers led through a year-long exploration with Google's Search UX Design Director — plus one launch calendar across three web properties.
Quick look — item details
warren·shop
Eleven months of research and design into production with Infosys and 28+ engineers, and a review cadence that kept skeptical executives inside the process.
Vendor ops
Quick look — item details
warren·shop
Owned Monaverse through implementation and platform testing to one launch, plus external motion, 3D-rendering and video vendors.
Visibility
Quick look — item details
warren·shop
Executive-level visualizations that kept a global campaign aligned on a cadence, with gated review holding quality across hundreds of assets.
Show me how work reaches Design and Research today and I'll tell you where the leaks are. That's the conversation I'd most like to have — and the fastest way to find out whether I'd be useful to you.