F3
15 days
Fractional Design Leadership
F3 - Design Governance and Assurance gives your named accountable person a route that turns a concern into a recorded decision. Accountability now sits with a name rather than a title, and the post carries the right to flag, recommend and escalate; then, when a release date arrives with a concern still open, somebody logs it, and the person closest to the customer absorbs it. With the Design Governance and Assurance module, we build the measurement and the escalation route, name an owner for each class of concern, and put the route inside the governance you already run.
Over six months, I hold that route open, and let it sit until it holds without me. You keep a governance vocabulary mapped to your regulator’s terms, a named owner and written decision right for each class of concern, and a board measure in the same format and cadence as your other risk measures.
On exactly that decision, MIT Sloan Management Review published an argument in June 2026, in a first-person essay by Adobe’s director of governance. The person with authority to halt a project either does not exist or operates as a paper tiger.
Real decision authority sits with somebody whose job is shipping. That is one practitioner’s argument in an essay, with no study and no measurement behind it, so I hold it as one.
My own fieldwork agrees, which is the only reason I repeat it.
Thirteen senior design and product leaders told me the same thing unprompted, across banking, insurance, fintech, media and software. One put it exactly: “legal, compliance, actuarial and the business product owner can; I cannot”.
Concerns fail to become decisions for four reasons, and I derived those four by working backwards through financial audit and information security, and then through NHS clinical governance.
Each of those three earned the authority to stop work, and each acquired something specific before it did. All three had to acquire vocabulary, ownership, escalation, and measurement, and design accountability is usually missing.
Adding a scanner to an organisation from no escalation route raises the number of flags while the number of things anyone can stop stays exactly where it was. The gap widens as the findings accumulate, and whoever sits closest to the customer absorbs the difference.
At a UK enterprise telco, I embedded three mandatory checkpoints inside existing project governance, and I declined to stand up anything parallel. A new forum is a new thing to ignore, and the escalation route has to live where decisions already get made.
When vocabulary, ownership, escalation and measurement fail together, the cost lands on particular people. In my fieldwork I found a lead signed off with stress for a month, and a designer who resigned after a regulatory crunch.
Contact centre managers elsewhere in the same fieldwork ran shadow spreadsheets for three months, because the system produced no number they could use. I call that the cost your people are quietly carrying.
I name who is carrying it, and I deliberately put no number on it. Organisations pay that cost for years without ever noticing they are paying it.
| Phase | Activities |
|---|---|
| Appetite | The four criteria tested upward as well as downward, before I build anything on top of them |
| Vocabulary | Language your organisation already uses, mapped to your regulator’s terms, with the class of quality that will not translate named out loud |
| Ownership | A named owner for each class of concern, with the decision right written down |
| Escalation | A route inside the governance you already run, with no parallel forum stood up |
| Measurement | A board measure in the same format and cadence as the other risk measures |
| Maintenance | The structure held while it beds in, and adjusted when you change your governance |
Appetite: everything after it depends on the answer. I run the four criteria upward and downward in the first month and establish that anyone above the seat wants quality for anyone, anything.
Vocabulary comes next, and it has to be language your organisation already uses. You know it has landed when people outside design repeat your concerns back accurately, in the words your regulator would recognise.
That test matters more than the document, because a vocabulary nobody repeats is a glossary.
Ownership and escalation belong together in the sequence. Ownership is real when you can ask somebody why a particular call went the way it did, and they answer without preparing.
Escalation is real when a concern reaches a decision forum without anyone having to volunteer to carry it there.
Measurement goes last, and it goes into the format that comes after risk and mistakes. The measure works when it appears on the page alongside the measures that already win arguments, at the same cadence, in front of the same people.
A measure on its own page loses to a measure on the risk page every time.
The mandate runs at one day a week, with a minimum of six months. The first month includes both the appetite test and the quality class that will turn it into harm.
After that, the work is maintenance: holding the structure while it beds in and adjusting it when you change your governance.
| The problem | What the structure settles |
|---|---|
| Somebody senior has been made accountable and has no structure behind the title to make it hold | A named owner for each class of concern, with the decision right written down and a route out of the concern |
| Concerns get raised, logged, and quietly absorbed | An escalation route inside the governance you already run, where a concern reaches a decision forum without anyone volunteering |
| A regulator, auditor or board committee has started asking how decisions were reached | A vocabulary mapped to your regulator’s terms, and a decision somebody can be asked to explain and does |
| You have findings from an audit or a tool and no route by which they become decisions | The route that turns a finding into a decision, so the flag count and the stop count move together |
All of that assumes there is executive appetite for quality to stop something, and this module fails when the four criteria above are failing and I hold ,seat. Where such appetite is missing, I will build good instrumentation, attach it to nothing, and I will know that inside six weeks.
I will say so, and you should stop paying me, which is why I run the criteria upward inside the first month. Six weeks is then the ceiling on what you have spent finding out.
The vocabulary criterion has a hard edge, and I would sooner tell you now than let you discover it later. In those same interviews, I found a class of quality that resists translation into harm so that nobody can govern it as risk at all, and so much of what your designers care about will never travel through this structure. No amount of building will change that. I name that set out loud in the first month.
Otherwise, people find it by watching their own concerns fail to register, and they conclude the structure is rigged.
This is one of four mandates under Fractional Design Leadership, which carries the offer and the fee. Read that page for the conflict rule I hold myself to.
If your problem is an accountable person with no structure behind the title, this is for you.
If your problem is that the evidence this structure would carry does not exist yet, then AI Scale Assessment is for you.
If the structure holds and the person who owns it is leaving, then Leadership Handover and Transition is for you.