Six mechanisms that made a design team impactful in the era of AI
When I present a year of design work to leadership, the deck has two parts. The first is the outcomes: what shipped and what it moved. The second is shorter and, to me, more important. It is what made those outcomes possible. Not the designers' talent, which was there before I arrived, but the mechanisms around it.
A mechanism is a fixed way of doing something that keeps working when nobody is paying attention. Good intentions do not survive a busy quarter. Mechanisms do.
They matter more now than when I put them in. A mock takes minutes and a prototype takes an afternoon, so any team can produce more screens than it could before, and a pile of screens has stopped being evidence of anything. What AI did not make cheaper is knowing which problem deserves the quarter, whether the work left the customer better off, and whether the people who fund the team can see that it did. Output is close to free. Impact is not. I have written about what that changed in how I lead; this piece is about the structure that was already there, and why it turned out to be the part that counts.
I run six mechanisms, in three groups: plan, execute, communicate. Each one replaced a specific way we used to fail, so I will describe each as the before and the after.
Plan
Ways of working: unpredictable to predictable
Before, design was a service desk. Work arrived from any direction, in any state, and each project found its own path from request to launch. Some had a kickoff. Some had a review. Which ones was a matter of luck, and so was the quality.
Now there is one agreed path, with a gate at every stage. Intake once a quarter, to align on priorities and estimate the work. A kickoff for every project, so the team shares one understanding of the users, the scope and the timeline. A deep dive early, to review direction while course correction is still cheap. A UX sign-off before engineering starts, which is the quality gate before the build. And fit and finish before release, which is the quality gate before launch, so that what was designed is what ships.
The line I use to explain why the gates sit where they do is Frank Lloyd Wright's: you can use an eraser on the drafting table or a sledgehammer on the construction site. Every gate is an eraser.
Intake: ad hoc to quarterly
Before, requests came in whenever someone thought of them, and the loudest one went first. Design was permanently interrupted and permanently behind.
Now every team submits its priorities once a quarter, with the business outcome each one serves and a design estimate, before any work starts. The quarter is planned against capacity, in the open, and the plan is the roadmap. Requests that arrive mid-quarter go through the same door and displace something visibly rather than silently. The share of unplanned work collapsed, and with it the feeling that design was always reacting.
Execute
Outcome-driven UX: output to outcome
Before, a project was done when the screens shipped. Nobody could say whether it had worked, so reviews were settled by seniority and confidence.
Now every project starts with the outcome, not the screen. Before any significant work begins, the team writes one sentence: whose life gets better, by what observable change, bounded in time, even when the hard situation occurs. The metrics follow from that sentence. A success metric defines the finish line. Progress metrics track the climb towards it. A problem-value metric prices what it costs to leave the problem unsolved. The threshold is agreed before the build, so it cannot be quietly moved once the results exist.
This is a way of executing, not a report written at the end. The outcome is decided first, and every decision after it is checked against it: which direction to pursue, which hard situation stays in scope, when the work is done. That matters more now that directions are cheap. A designer can explore more of them than we ever had time for when every screen was drawn from scratch, and without a written outcome, more options only means more opinions. With one, the choice is short.
Design is judged by what moved, not by what shipped. I have written about the framework behind this separately, because it deserves its own piece. The short version: use metrics count actions, outcome metrics verify the action left the customer better off, and only the second kind gets to be the judge.
Critique: siloed to a shared bar
Before, one designer held one view of one project, and consistency across the product was whatever happened to emerge. Two screens built in the same month by two people could feel like two companies.
Now the team reviews work in progress together, every week, against the bar. A designer brings a flow, states what it is, why it exists, the risk, and what they need from the room. Consistency is decided together, once, rather than per project by whoever is nearest. The bar rises because everyone sees everyone else's work, and because the review is scheduled, it happens even in the weeks when nobody feels like it.
Communicate
The newsletter and the leadership deck: ask to broadcast
Before, if you wanted to know what design was doing, you asked a designer, who stopped working to tell you. Status lived in people's heads and in scattered threads, and every stakeholder had a slightly different picture. Leadership had the thinnest picture of all. It saw launches, not what the launches did.
Now design tells the organisation before anyone asks, on two rhythms.
Every two weeks, one note goes to the whole organisation: every project in flight, its goal, the latest update, its status, its date, its owner, its health. Progress, decisions, what is next. It is deliberately lightweight, and it is the mechanism that changed how design was treated. Nobody has to ask, so nobody interrupts, and people started arriving at conversations already informed. A function that tells you what it is doing reads as a function that runs itself.
The leadership deck I opened with is the same idea pointed upward. It is not a tour of screens. It reports the outcomes, each one against the threshold we published before the build, and then the mechanisms that produced them. Leadership prefers to read progress as outcomes delivered rather than percentages moved, so each result is stated as a sentence about a customer: what they can do now that they could not do before, and what has stopped happening to them.
The deck and the outcome work are two halves of one thing. Outcome-driven UX is how the work is executed. The deck is how it is communicated. Each fails without the other. Outcomes that nobody hears about do not change how a team is treated, and a deck with no outcomes behind it is a portfolio. When every function can show a great deal of output, a portfolio proves nothing. A short list of things that moved, with the numbers agreed in advance, is hard to argue with.
People: undefined to defined
Before, there were no written expectations for any role. Designers did not know what growth looked like, and neither, if I am honest, did their managers. Promotion conversations were about feelings.
Now there is a written expectations document for every level, from associate to principal: what you own, what a plan you build must answer, how you work with partners, what judgment you are expected to show. Weekly one-to-ones are about growth against those expectations, not status, because status is in the newsletter. When someone is promoted, everyone can see why, and everyone else can see what would get them there.
What the six have in common
None of them is clever, and none of them is about AI. Every one of them replaces a dependence on a person with a dependence on a rhythm: a quarter, a week, a fortnight, a gate. That is the whole trick. Talented people in an unpredictable system produce unpredictable results and burn out doing it. The same people inside these six mechanisms produce outcomes on a schedule, and they get to spend their energy on the work rather than on defending it.
Predictability was the first thing the mechanisms bought. Impact is what the predictability was for. AI changed the speed of the making. It did not change what makes the making count, and a team with fast tools and no mechanisms only produces unpredictable results faster. The plan decides where the speed is spent. The outcome decides whether it was worth spending. The newsletter and the deck make sure the organisation knows.
If you lead a design team and you are choosing where to start, start with the newsletter. It costs two paragraphs every two weeks, and it buys you the credibility to install the other five.