5 guests 5 episodes 3,877 words

Who Holds the Pen? The Centralized vs. Distributed Battle for Design Taste

Should the design quality bar be enforced by a small group of taste-keepers at the top, or distributed across every PM, designer, and engineer?

At Shopify, nothing ships without Glen Coates personally okay-to'ing it on behalf of Tobi Lütke. At Snap, Evan Spiegel overruled his entire leadership team to declare the company a "camera company" and nobody else at the org could even articulate why the call was right. At Airbnb, Brian Chesky renamed every product manager to "product marketing manager" so there would be no ambiguity about who actually owned the product. And then there is Linear, where one product manager total runs an entire 50-person org, and Karri Saarinen insists craft is held not by him but by every engineer the company has been willing to wait months to hire.

These are not stylistic preferences. They are opposing theories about where taste actually lives inside a company — and about what happens to quality when an organization grows past the point where a single founder can still feel every pixel. Archie Abrams, Gaurav Misra, Sanchan Saxena, Karri Saarinen, and Katie Dill have all had front-row seats at companies that are famous for design craft, and they fundamentally disagree about whether that craft is a thing you gate or a thing you distribute.

You are running product or design at a company that has crossed the point where the founder can no longer personally review every screen. Do you concentrate the quality bar in a tiny group of deputized taste-keepers — the founder plus two or three people empowered to speak for them — and funnel every release through that group? Or do you push the responsibility for craft into every PM, designer, and engineer, using hiring and rituals instead of approvals to hold the line?

The 3 Positions

1 / 3

Concentrate the Bar — Taste Is Rare and Must Be Held by a Few

Archie Abrams · Gaurav Misra · Sanchan Saxena

Archie Abrams — VP Product and Head of Growth at Shopify, running a 600-person growth org under Tobi Lütke and Glen Coates. His description of how taste gets operationalized at one of the most famously founder-driven product companies is first-hand, and his endorsement of centralized taste for core product is credible precisely because it runs against the metrics-driven instincts of his own growth discipline.

Archie's argument starts from a concession opponents rarely acknowledge: design quality is inherently subjective. The moment you try to run a company on "taste and intuition" instead of metrics, you create a dangerous squishiness — everyone has their own opinion of good, and without a hard anchor the aggregate drifts into incoherence. Shopify's answer is to make subjectivity tolerable by concentrating it in a very small number of people who genuinely hold what the bar is.

"It's subjective, but it's objective in the sense that it's kind of a small number of people who really hold what that bar is and needs to be. I think if it's just subjective, let's just ship what we want without a couple people really holding that quality and that taste bar, that's where things go really sideways."
Archie Abrams
▶ 00:51:00

Tobi sets the 100-year vision. Glen is the operational enforcer — "founder mode, but not as a founder," as Archie puts it — a deputy whose taste is close enough to Tobi's that Tobi doesn't have to touch every decision. Archie, Glen, and Carl do "okay-to" approvals for their respective areas, and no project can ship without one. Every six weeks the entire leadership team meets in person and walks through every project, video by video, Figma by Figma, to hash out disagreements.

"Glen, who's heads of core product, I mean one of the things that's so impressive about Glen and that core team is they go incredibly deep into every single release that is shipped. And so you do have a central eye on the quality and how it all fits together."
Archie Abrams
▶ 00:50:39

"Every single project that goes out has a few minute video with Figmas and everything, and everything that shipped. Needs to be okay-toed, so approved by the group lead. There's nothing that can ship without that okay-to approval. And that okay-to approval has to be Glen, Carl, myself with different groups."
Archie Abrams
▶ 00:52:12

The structural claim is that the "okay-to" gate is a forcing function — a single ritual that keeps the entire org anchored to one taste even as it scales past hundreds of PMs. The catch, which Archie is careful to name, is that it only works if the people at the top actually have exceptional taste. "It's rare that you're a Tobi or a Brian Chesky or Elon."

Gaurav Misra — CEO and co-founder of Captions (10M+ users, $100M+ raised) and a former design engineering lead at Snap who reported close to Evan Spiegel. One of very few operators who saw Spiegel's product process from the inside, and who is now consciously trying to port that model to his own company.

Gaurav's take is the purest version of the centralized thesis: deep user taste cannot be distributed because it cannot even be verbalized. The most unique thing about Snap, in his telling, was that Evan had an understanding of the user that nobody else in the company could match — not even the rest of the leadership team — and every time they overruled him the product suffered for it.

"I do remember at Snap for example, I think one of the most unique things about Snap and the CEO Evan Spiegel was that he had an unmatched understanding of the user. I think years and years and years of the company's existence past, almost a decade. And nobody understood the user like he did. He would come up with ideas that everybody would disagree with and we would launch them and there would be hits, just hits after hits. And nobody would understand why."
Gaurav Misra
▶ 00:33:26

The canonical example is Snap declaring itself a "camera company" — a call the entire leadership team disagreed with, that outsiders mocked, and that turned out to be the single most defensible decision the company ever made. Opening directly to the camera meant Instagram could never copy it without tanking their own engagement metrics on day one.

"I think there was a point at which Snap declared that they're a camera company and a lot of people laugh at them... But I think at the core of it was this idea that Snapchat opens to the camera and that was actually the differentiator. That actually that small decision was holding the entire company against all competition."
Gaurav Misra
▶ 00:34:07

The structural punchline is that this kind of intuition is impossible to hand off. Gaurav can't even load his own taste into an LLM context window for his own PMs to use — and if he can't verbalize it, he certainly can't distribute it to a committee. At Captions this shows up as a deliberate "secret roadmap" of founder-curated bets, insulated from the distributed feedback machinery, because distributed feedback always converges toward the obvious.

"The problem is context. And I think just the context of understanding the user, the use case, it's so abstract. Even right now, I feel like I understand our users obviously, but I can't exactly verbalize why that is or how that is... And I spend a lot of time with PMs and designers imparting anything that I understand and I've learned over the many years I've been working on this, how do I impart this to them? But then it's a challenge because I can't even verbalize it myself."
Gaurav Misra
▶ 00:32:25

Sanchan Saxena — VP of Product at Coinbase, previously Head of Product at Airbnb where he scaled the PM org from ~30 to ~200 through hypergrowth, IPO prep, and COVID. Worked directly under Brian Chesky through the crisis and has a rare front-row seat on the Chesky model at both its best and its most stress-tested.

Sanchan's argument is experiential and unapologetically founder-centric. Airbnb was — and in his view had to be — a founder-driven company, and the reason it survived COVID while other travel companies collapsed is that Brian was personally on the front line making product, business, and emotional decisions himself rather than delegating them to a committee or a layer of PMs.

"I learned all things about building great products, building great businesses, building great companies from founders like Kevin Systrom and Brian Chesky, Brian Armstrong and others... I got I think a masterclass from Chesky in how to build products by watching him act, behave, and make decisions as well."
Sanchan Saxena
▶ 00:13:39

The mechanism of taste transmission in Sanchan's telling is proximity, not process. You learn the craft by watching one exceptional person decide and absorbing the principles directly. That is why Brian stayed deep in the trenches — and also why he eventually renamed every Airbnb PM to "product marketing manager," a move widely read as controversial but which Sanchan interprets as a deliberate reassertion that the founder and the design org, not the PMs, owned the product.

"One of the things about Airbnb... it's a very founder-driven company. While there are leaders over there, the founder is deeply in the trenches. He really understands the customer. He's not this executive who's so far removed from reality that they can't actually create the content. Brian was living, breathing that exercise."
Sanchan Saxena
▶ 00:21:48

"The secret sauce of that in my opinion is the founder. And if the employees do not believe in the founder, if the employees do not believe that the founder can make things happen as they talk about... those companies go down in flames. The second thing I would say is the founder needs to be at the front line. And Brian Chesky was on the front line."
Sanchan Saxena
▶ 00:26:17

The implicit claim for the taste-bar debate is that when the product is a real-world experience — a stay, a trip, a memory — no committee will make the trade-offs required to hold craft. It needs one person who feels the whole thing.

When This Applies

when you have a genuinely exceptional founder whose product intuition is stronger than anything the rest of the org could produce by committee, when the product depends on non-obvious bets that distributed feedback will average away, and when you can tolerate the throughput cost of a tiny approval group. Snap under Spiegel, Airbnb under Chesky, and Shopify under Tobi and Glen all fit here.

Distribute the Bar — Craft Belongs to Everyone, Enforced by Hiring

Karri Saarinen

Karri Saarinen — Co-founder and CEO of Linear, previously principal designer at Airbnb and founding designer at Coinbase. A designer who lived inside the Chesky-centralized environment and deliberately built his own company the other way.

Karri is the most interesting voice in this debate because he has run the experiment on himself. He absorbed the Airbnb model from the inside and concluded that its bottleneck was structural, not solvable. Linear has exactly one product manager — the head of product. Engineers and designers are expected to do the PM work themselves: scoping, writing specs, running the project, making the taste calls. Specialization, in his view, is efficient in a Ford-factory sense but destroys craft, because craft requires every person in the loop to be thinking about the whole.

"We just think that in order to build quality things or build things in a certain way it's actually better if people actually also spend some time thinking about things and not just executing."
Karri Saarinen
▶ 00:19:35

The forcing function that makes this work is not a gate; it is hiring. Linear doesn't just interview engineers for engineering skill. It interviews them for product judgment — can this person form an opinion about how the product should work and defend it? If the answer is no, they don't get hired, no matter how technically strong they are. That filter is what lets Karri trust craft to the team rather than gating it at the top.

"You cannot really interview engineers only for the engineering skills. You also have to interview them for the product skills... We just look for does this person have opinions about products and how they work? Can they form opinions and can they use their own judgment at times?"
Karri Saarinen
▶ 00:21:32

Karri himself is not absent from craft decisions — he still runs weekly design reviews, co-founders sponsor projects, and he personally tries every feature before it ships. But his role is coach and last-line reviewer, not sole gatekeeper. The telling detail is how he describes catching bugs: not by reading Figmas in an approval meeting but by actually using the product.

"Before we are launching it, I might just go in and try it out and try the different states and click it around. And sometimes I find things, like we were building this threading to comments and then when it looked all good in the demos... and then I went to try it and try different lengths of messages and stuff and then I start to see, oh, sometimes the animations are kind of janky or it's just off."
Karri Saarinen
▶ 00:34:27

The hidden assumption he is explicit about: you have to be willing to hire much more slowly and reject people with great skills but no product opinions. Most companies aren't.

When This Applies

when you are willing to hire very slowly, reject technically strong candidates who lack product opinions, and build the rituals (weekly reviews, paid work trials, sponsored projects) that keep distributed judgment calibrated. Linear is the canonical case, and almost nobody else has actually paid the hiring cost.

Craft Is a Cross-Functional Vocabulary, Not a Franchise

Katie Dill

Katie Dill — Head of Design at Stripe, previously head of design at Airbnb Experiences and at Lyft. One of the most senior design leaders in fintech and marketplaces, with direct experience running design inside a Chesky-led org and then scaling it in the very different, design-distributed cultures of Lyft and Stripe.

Katie's position sits between the two camps but leans hard toward distribution. Her framing is that the whole "business goals vs. design goals" dichotomy is itself the bug. If the entire cross-functional team is aligned on what "great" means, then craft becomes a shared vocabulary rather than the private franchise of a design department or a founder. The engineer cares about the checkout error state. The PM cares about the emotional texture of the flow. The designer cares about conversion. Nobody is defending craft against anybody.

"A pet peeve of mine is this way of talking about things as there's business goals and there's design goals. Because I think maybe the first conversation one should have is that, 'What are we trying to build towards?' And I would think that folks that want to create really impactful products, they want to create quality products... because they know that will build a stronger business in the long run."
Katie Dill
▶ 00:19:13

Design, in her reframing, is not a discipline — it is intentionality. And intentionality is something any function can bring.

"At the core, design is simply intention. You're bringing intentionality to the decisions that you make in thinking about who is this thing for... These are some of the intentional decisions one could make, whether or not they're a designer, an engineer, the product person, any old function can put that intentionality."
Katie Dill
▶ 01:19:24

The commercial receipt for this thesis is Stripe checkout. By pushing craft thinking into every person on the flow — Stripe famously operates with a very lean PM layer, pushing ownership onto engineering and design ICs — the team measured a 10.5% revenue lift from the older checkout to the newer one, without any central taste-keeper banging the table.

"By improving the quality of the checkout experience through details small and large, we have seen a 10.5% increase in business' revenue from an older form of checkout to a newer form of checkout. And those little details matter to have such a material impact on one's revenue."
Katie Dill
▶ 00:17:14

The concession hiding inside the model: distributed craft still requires extremely senior design leadership setting the vocabulary. Otherwise "everyone owns it" becomes "no one owns it."

When This Applies

when the founder has the taste but cannot personally touch every release anymore, and you can identify a small number of people whose judgment is genuinely interchangeable with the founder's. Shopify's Glen Coates "founder mode but not as a founder" model is the cleanest example — and it is much rarer than founders assume.

Evidence from the Archive

Shopify

At Shopify, nothing ships in core product without an 'okay-to' approval from Glen, Carl, or Archie — a three-person gate for a 600-person org

A tiny central group holds the taste bar for every release; Glen Coates reviews every single one in depth before it ships

Archie Abrams · VP Product, Head of Growth at Shopify — works directly with Tobi Lütke and Glen Coates ▶ 00:51:00

Snap

Evan Spiegel proposed ideas the entire Snap leadership team disagreed with — they launched anyway and were hits nobody could explain

Gaurav Misra watched Spiegel make product calls from a position nobody else at the company could match, including the 'camera company' decision that held Snap against Instagram forever

Gaurav Misra · CEO and co-founder of Captions; previously design engineering lead at Snap ▶ 00:33:26

Airbnb

Brian Chesky renamed every Airbnb PM to 'Product Marketing Manager' to reassert that the founder and design organization owned taste

Sanchan Saxena scaled Airbnb's PM org from ~30 to ~200 under Chesky and learned the taste bar only survives if the founder is personally on the front line, not delegating

Sanchan Saxena · VP of Product at Coinbase; former Head of Product scaling at Airbnb ▶ 00:26:17

Linear

Linear has exactly one product manager for the entire company — engineers and designers are hired for product judgment and own craft end-to-end

Karri Saarinen deliberately built Linear the opposite of Chesky's Airbnb: distributed taste via extreme hiring, with the founder as backstop not gatekeeper

Karri Saarinen · Co-founder and CEO of Linear; previously principal designer at Airbnb, founding designer at Coinbase ▶ 00:21:32

Stripe

Stripe measured a 10.5% revenue lift on checkout from distributed craft investment across PM, eng, and design — not a central taste-keeper banging the table

Katie Dill runs design at Stripe with deliberately few PMs and pushes craft responsibility into engineering ICs; the commercial payoff shows up in checkout flow metrics

Katie Dill · Head of Design at Stripe; previously Airbnb Experiences and Lyft ▶ 00:17:14

The Synthesis

The surface debate here — centralize or distribute — is actually a debate about three things hiding underneath it.

Taste's Verbal Ceiling

Can you even articulate your taste?

Gate vs. Filter

Do you hold the bar by approving or by hiring?

The Deputy Problem

Is your Glen actually a Glen?

The Portable Model Myth

What's the prerequisite you're missing?

Replicate or Defer

Is the org trained to copy taste or defer to it?

If taste is a pre-verbal intuition like Gaurav Misra describes at Snap — something Evan Spiegel couldn't explain but couldn't be overruled on — centralization isn't a choice, it's physics. You can't distribute what you can't put into words. The moment you can articulate the bar, the centralized model loses its only structural justification.

Shopify's okay-to gate and Linear's hiring filter do the same job — they stop the bar from drifting. One concentrates authority in Glen Coates; the other distributes it by rejecting technically strong candidates without product opinions. Pick your lever, but know you need one: craft without either collapses into incoherent shipping.

Centralized-through-deputy only works when the deputy's judgment is genuinely interchangeable with the founder's — 'founder mode, but not as a founder.' This is much rarer than founders assume. Copy Shopify's gate without a Tobi-and-Glen pair and you entrench mediocre taste at the top with all the ceremony of excellence.

None of these systems are portable; the prerequisites are. An exceptional founder, a deputy-grade executor, a ruthless hiring bar, or extremely senior design leadership — every model depends on a scarce resource process cannot manufacture. Karri's distributed model without Linear's hiring cost produces chaos, not craft.

The founder always holds the bar — Karri personally catches janky animations in Linear's features before they ship. The real question is whether the rest of the org is being trained to replicate that judgment or trained to wait for the founder's approval. One builds capacity; the other builds learned helplessness.

Which Approach Fits You?

Answer 3 questions about your situation. We'll match you to the right approach.

Question 1

How would you describe the founder's relationship to product detail at your company today?

Question 2

What is your hiring speed and selectivity actually like?

Question 3

What does your product surface look like?

Notable Absences

The Bottom Line

The ugly conclusion for most product leaders: if you copy Shopify's gate without a Tobi and a Glen, you will entrench mediocre taste at the top. If you copy Linear's distribution without its hiring filter, you will get incoherent shipping. The model is not portable. The prerequisite is.

The non-obvious insight is that every one of these voices, including the distributed ones, is still describing a founder-anchored system. Karri catches janky animations personally. Katie Dill is a very senior design leader whose vocabulary the team is calibrating against. The real question isn't whether the founder holds the taste bar — she always does — but whether the rest of the org is being trained to replicate that taste or trained to defer to it. Shopify's okay-to gate produces reliable coherence but risks learned helplessness. Linear's hiring bar produces distributed judgment but is extremely slow and has not yet been tested at 500 people. Neither side has actually won the scale test. The companies that have tried to graduate from centralized to distributed — and most of them try, eventually — describe the transition as one of the hardest things they have ever done.

Sources

Archie Abrams — "Breaking the rules of growth: Why Shopify bans KPIs, optimizes for churn, prioritizes intuition, and builds toward a 100-year vision | Archie Abrams (VP Product, Head of Growth at Shopify)" — Lenny's Podcast, November 7, 2024
Gaurav Misra — "How to win in the AI era: Ship a feature every week, embrace technical debt, ruthlessly cut scope, and create magic your competitors can't copy | Gaurav Misra (CEO and co-founder of Captions)" — Lenny's Podcast, March 27, 2025
Sanchan Saxena — "Sanchan Saxena (VP of Product at Coinbase) on the inside story of how Airbnb made it through Covid; what he’s learned from Brian Chesky, Brian Armstrong, and Kevin Systrom; much more" — Lenny's Podcast, July 5, 2022
Karri Saarinen — "Inside Linear: Building with taste, craft, and focus | Karri Saarinen (co-founder, designer, CEO)" — Lenny's Podcast, October 8, 2023
Katie Dill — "Building beautiful products with Stripe’s Head of Design | Katie Dill (Stripe, Airbnb, Lyft)" — Lenny's Podcast, October 15, 2023