Every design leader I have watched fail has failed in the same way, and it was never on craft or taste. They failed because they believed the function they led was the point.
It isn't. Design exists to move the business. So does engineering, so does marketing, so does the function you happen to run. That sentence gets read as cynicism by people who have never had to defend a headcount, and as obvious by anyone who has. It is neither. It is the single most useful thing I know about doing this job, and almost nobody says it out loud to the people who most need to hear it.
I want to make the case properly, because the version of this argument you usually encounter is a bad one.
The bad version
It says design should stop whining, learn to read a balance sheet, and do what it's told. That version produces order-takers. It produces teams that ship whatever the loudest stakeholder asked for and never say no. Then get cut first in a downturn, because they made themselves interchangeable with a contractor.
Subordinate does not mean servile. A cardiologist is subordinate to the patient's survival. That doesn't make them a technician. It makes them accountable to something other than their own preferences.
Leaders who insist design deserves a seat at the table almost never get one. Leaders who understand they are there to move the business get handed the table.
What accepting it actually buys you
When I joined Honor we were building for two customers at once. The families who needed home care, and the Care Pros who delivered it. That was our term for the caregivers, and the naming was not incidental. Those are different products with different economics, and the temptation was to treat the caregiver-facing tools as a moral obligation, the right thing to do, funded out of goodwill.
That framing would have killed them in the first budget crunch. Goodwill is the first line cut. What saved them was an economic argument. Better tools meant Care Pros stayed. Retention drove supply, supply drove the marketplace, and the marketplace was the company. Higher pay and better software weren't charity. They were the business model. Same work, same values, completely different durability, because it was tied to something the company could not cut without cutting itself.
At Lyft the transition was from a car-based rideshare app to something multimodal. Passenger, transit, bikes and scooters, marketplace, AV. You can frame that as a design opportunity, and it is a gorgeous one. But it got resourced because a rider who uses three modes is worth more than a rider who uses one. And we needed a defensible position in cities turning hostile to cars. Our design work was better because it was pointed at that, not despite it.
At Grammarly we moved from a browser extension to native Mac and Windows apps. It looked like an interface project. It was really a question about distribution, wearing an interface costume. Where does writing actually happen, and can we be there? The company was hiring in the US, Canada and Ukraine. Answering the distribution question was a hiring problem, an onboarding problem and a culture problem long before it was a design problem.
The test
Here is how you find out where you stand. Take the thing your team is proudest of this quarter. Now say what it did for the company, without using the words quality, craft, consistency or experience:
We did this, which caused ______, which the company needed because ______.
If you can't complete it, that work is undefended. It may still be excellent. It may still be the right thing. But it is undefended, and undefended work goes first. Not because your executives are philistines. Because they are allocating finite money against claims that are defended.
Most design leaders I've interviewed cannot finish that sentence about their own best work. They can talk beautifully about how they work. They cannot say what the work did, or by what route it did it.
Where the line actually is
So what do you refuse?
You refuse the thing that wins this quarter and costs you the next four. Dark patterns are the obvious case, but they're rare enough to be a distraction. Common cases are duller. A shortcut that adds permanent upkeep. A feature that serves the loudest customer and confuses the other thirty thousand. A reorg that looks efficient on a slide and severs the one relationship making delivery work.
Notice that all of those refusals are also business arguments, which I feel is the whole point. You are not appealing to a separate authority called "design." You are saying: this is a bad trade by your own measure, and here is how it goes wrong. That argument can be won. "This isn't good design" cannot, because it asks your CFO to defer to a taste they do not share and cannot audit.
Hard cases are the ones where you're right and lose anyway. I'm not sure there is a technique for those. I have lost plenty. What I've learned is to lose in a way that leaves the argument standing: write it down, name the mechanism, date it. Half the time the company comes back to it in nine months. Half the time you were wrong, and you have a record of how.
Why nobody says this to you
Because the conference-talk version is more flattering. There is a whole genre of design-leadership content built on the premise that the discipline is constantly undervalued and the fix is more advocacy. It sells well. It is also, in my experience, close to backwards. That undervaluing is real. It is also a symptom. It happens because the function keeps arguing in a language the business does not pay for.
I don't think you should love this. Notice that the people who internalized it early got more scope and protected more of their teams through cuts. And, this is the part that surprised me, shipped better work. A constraint you can trace to something real produces better work than a constraint that is just somebody's preference.
Subordinate to the business is the whole job, not a smaller one. Craft is what you do with it.
Where I took this too far
I have been arguing for this frame for years, so I should say where applying it made me worse at the job.
There was a stretch in the middle of my time running larger organizations when I could complete the mechanism sentence about anything. I had stopped asking whether the work was any good. Every project had a defensible commercial story. Every review ended with a metric. And the quality of what we were shipping had drifted somewhere I was no longer looking. Nobody flagged it, because the numbers were fine and I had trained the room to speak to me in numbers.
The frame had become a condition of starting rather than a way of defending. At Honor we had committed to the product working over a landline as well as a smartphone, and nothing measurable sat next to that decision for years. Required at the point of commitment, it would never have survived.
What I would say to a younger version of me: the mechanism is how you defend the work, and it is a poor instrument for choosing it. Some of what a design organization should be doing has no commercial story available yet. Require one at the point of commitment and you will kill the early version of everything that later turns out to matter. Holding both is the discipline, and it is harder than either. Be explicit with your team about which mode a conversation is in.
One thing the frame cannot survive: a company whose time horizon is short enough that "the business" stops being a proxy for anything worth building. Optimize faithfully for that and you will build something worthless. Which is a reason to choose your company carefully rather than to reject the frame, though I would want it said out loud.
Subordinate to the business is still the whole job. It is a claim about accountability, and I spent a period treating it as a claim about taste.
I've led design and product at Honor, Lyft, Grammarly and Pitch, and today I build at Noord.
© 2026 Renato Valdés-Olmos