The Signal Noise Playbook
Before Sedna, I built the first shared production system at Signal Noise, a data design consultancy inside The Economist Group. There was a production team and a head of production, but no common way of working: every producer ran their projects differently, so quality and predictability varied. Same shape as everywhere I've worked, I turned that into a discrete, repeatable function. This is where the four-pillar systems approach behind the rest of this site started.
A production team, but no shared system.
Signal Noise had a production team and a head of production, but no common way of working. Every project was run differently and every producer ran theirs their own way, so quality and predictability depended on who was in the room. Over eighteen months I redefined how the studio executed projects and packaged it as a reusable system: the Signal Noise Playbook. We built it the way you would build a design system, taking inspiration from IBM's Carbon, but for production rather than UI.
Decide before you commit.
I redefined the project phases and separated the statement of work from the scope of work, fixing scope only at the end of a Discovery phase, as a Method Statement. That gave teams time to find the real need before prescribing a solution.
- 01
A scoping and estimation tool: define the skills needed, estimate effort, prioritise features, and flag what could be deprioritised to protect margin.
- 02
A cause-effect framework that made the cost of the three most common delays explicit (new stakeholders, missed reviews, missed milestones), so client responsibilities were clear before things went wrong.
Delivery aligned to business objectives.
A single thirty-day objective-setting session where senior leadership fed priorities to the team leads. From those priorities, a Kanban of "plays", each mapped to the business objective it served. The plays were designed to be taught and coached outside any single project, so quality did not depend on who happened to be staffed.
A system that learns.
I added retrospectives to sprints and project closes, then went a step further: a filterable knowledge database, so lessons and blockers from past projects informed new ones of a similar type. A memory for production.
Bigger than any one person.
Critical knowledge used to leave every time someone moved on. The playbook lived in Notion as a single source of truth so it outlived any individual who used it. It became the default onboarding tool, using plays and Loom videos so new joiners could ramp at their own pace. And it was team-owned: each play could be edited by the teams using it, with a changelog, because a production system is only as good as the belief each team has in it.
The result.
Teams also reported more collaboration and less time in meetings.
We needed to codify how we work internally and with clients in a way that could continually evolve. The playbook is about working with intention and continual self-reflection.
A clear production methodology, responsive to the needs and learnings of the team, gives everyone a sense of security that there's a way to do a given thing, that it's been thought about, and that things won't be left to chance.
Signal Noise is where the four-pillar approach behind this site started: Discovery, Delivery, Cadence, Learning. The AI toolkit is the same thinking, now automated.
This was a data design consultancy, not a product company. The domain differs; the discipline of building a delivery function from scratch is the same. Profit-margin, planning-time and late-delivery figures are from my own records of the period; the original published case study reported qualitative outcomes.