TURNING TASTE INTO AN OPERATING SYSTEM

WHAT HAPPENS WHEN “I KNOW IT WHEN I SEE IT” IS NO LONGER ENOUGH?

Of all my childhood memories, the ones that stuck with me most were rather mundane on the surface - like the walks I'd take with my mom around the block. We grew up in a forested neighborhood with a mix of 1970s ranch homes, New England-style cottages, wraparound porches and stately suburban Georgians.

Looking back, those walks were really full-blown sessions in architectural criticism. We weren't exactly malicious, but like most Southerners in the heat of August, we certainly had opinions. A beautiful old house with the wrong shutters. A front door painted exactly the right shade of green. A gaudy mailbox that inexplicably ruined the whole thing.

Someone who had clearly spent a fortune renovating and somehow made the place worse. Someone else who had done almost nothing and gotten it exactly right.

It became a ritual between us. We would walk, look, compare and talk about why one thing worked and another didn't. I didn't have the language for it at the time, but those walks were teaching me how to see. More importantly, I was learning that taste wasn't simply about liking something. It was about noticing the small decisions that made something feel coherent, distinctive, beautiful, completely off or almost perfect.

Most of us develop taste this way. Not through a rubric or a course on aesthetics, but through repeated exposure. We look at things. Make things. Travel. Read. Compare. We try our hand at something and get it terribly wrong. We change our minds and start to notice what other people miss. Over time, those encounters accumulate into a kind of internal recognition system.

Eventually, you can walk into a room, look at a piece of work or encounter an idea and think: damn, that's good. Or, more dangerously: it's not quite there.

For a long time, that can be enough. If you're the person making the thing - or simply deciding whether you want it - instinct is wonderfully efficient. The trouble begins when that judgment has to travel beyond you.

I ran into this problem at a much bigger scale while working on Airbnb Experiences. As curators, we reviewed experience concepts from cities around the world and had to decide whether they aligned with our product vision, brand story and quality bar. The decisions often sounded simple until you tried to explain them.

A host might be genuinely credible while the activity itself felt generic. A venue might be beautiful but incidental to what was happening there. An experience might be flawlessly produced but leave guests feeling passive. Something might be wonderfully niche yet still not feel compelling enough for someone to spend three hours trekking outside the city to do it.

Over time, you could feel the difference between a concept that nailed it and one that faltered. But "I know it when I see it" becomes a fairly useless operating principle when eight curators, dozens of reviewers and teams across different continents all need to know what great looks like. And when it comes to giving creative feedback at scale, "not quite there" just won't cut it.

The challenge wasn't to reduce qualitative taste to a formula. It was to figure out how judgment could travel without becoming generic. How do you preserve the ability to sense something without turning it into a checklist? How do you create consistency without defaulting to sameness? How do you establish a shared understanding of quality that can work in Paris, Tokyo and Mexico City without quietly asking all three places to behave like one another?

The more I worked on this challenge, the more I began to think of taste less as an attribute someone possesses and more as an operating system other people can learn to work with.

Developing an operating system for taste begins with signal

Before you can scale a creative standard, someone has to notice what matters. The things we give weight to in the act of noticing are signals: qualitative aspects of something that capture our attention. For an experience, the signals we care about might be authenticity, a host’s expertise, an activity’s relevance, the way guests participation or a meaningful relationship between the activity and the place it happens.

Is this activity specific to this cultural context? Is this host genuinely expert in their field? What evidence points back to their mastery? Is this experience quintessential to this place, or could the same thing happen anywhere?

These questions matter because they provide the type of information we need to differentiate something “meh” from something genuinely great.

We live in a moment of information abundance. We're surrounded by reviews, trend reports, market research, customer feedback and increasingly powerful AI systems capable of spitting out and synthesizing astonishing amounts of data. What is scarce is often knowing what information, or signal, deserves our attention inside the noise.

Why does one host feel magnetic while another person with almost identical credentials does not? Why does one restaurant feel deeply rooted in place while another could have been dropped into any fashionable neighborhood in the world? Why does an experience that looks less polished somehow feel more alive? Which quirky consumer preference is an anomaly, and which is an early signal of where culture is going?

Identifying those signals requires context, comparison and judgment - - and sometimes the uncomfortable admission that the thing performing best on paper is not necessarily the thing worth protecting.

Signal is useless without translation

Once you recognize what matters, you have to find language for it that can survive outside your own head. Creative organizations are full of perfectly intelligent people saying things like make it more brand aligned, it needs more magic, this doesn't feel elevated enough, make it culturally relevant, or my personal favorite, we'll know it when we see it. These statements may describe a real sense of intuition, they don’t give someone else much clarity about what to do next.

The work of translation requires getting underneath your reaction to something and articulating why something matters, what good looks like in practice and how another person might move the work closer to it.

Take place. Imagine you're reviewing an experience in Lisbon where a host is a local fadista with thirty years of experience performing fado at some of the city's most incredible venues. The problem is that the only location they can secure for a night of music and petiscos is the cramped community room of a local church: fluorescent lights, folding chairs, not exactly the atmosphere the music deserves.

You could tell the host they need a "more suitable venue." True, perhaps, but not especially useful to someone dealing with real operating constraints. Better feedback explains what suitability looks like and gives them a principle they can adapt: the best experiences happen in places that enrich the activity rather than simply contain it.

Then you can ask a real question: This experience sounds great, but I’m worried that the venue might be a bit small for this activity. Do you have access to another venue where your performance might feel more meaningful and comfortable for guests?

Suddenly an instinct becomes something another person can reason with. The activity works but the venue needs to change for a specific reason. A host can respond to it. A team can debate it. At scale, an AI model can surface evidence to reason against it.

Two people might interpret the same principle that defines a signal differently without either one necessarily being wrong, and that’s by design. The effective translation of taste creates shared language without pretending judgment has become objective.

The infrastructure of taste is multidimensional

A creative director can establish a point of view. A creative team can champion it through their decisions. But when taste has to scale across a global organization — much less a public marketplace - - eventually you need artifacts, systems and infrastructure that help it move beyond the people who first defined it.

That infrastructure might begin with clear creative criteria and expand into a library of examples, recurring rituals for calibration and critique, training, prompts, decision trees and tools that help people apply principles under pressure.

At Airbnb, I became increasingly interested in how AI could become part of this infrastructure. Not because I wanted a model making consequential creative decisions instead of curators, but because so much of the work surrounding judgment could suddenly become faster and richer at scale.

We had already spent months identifying what good looked like, translating that into specific language and collecting examples for tricky edge cases around the world. I used those inputs to build AI-assisted tools that could retrieve hyperlocal context, compare an idea against quality criteria, surface contradictions and point a curator toward where human attention mattered most before they made the call.

The goal was not to outsource taste. It was to give human judgment more capacity.

That matters when you are making dozens of nuanced creative decisions in a matter of hours. Even the most experienced curator loses context eventually. After enough reviews, unusual cases start to look familiar, subtle distinctions become harder to hold in your head, and the temptation to default to the most obvious interpretation of a standard gets stronger.

Now with the click of a button, a curator could leverage AI to draw on hundreds of prior edge cases, real-world examples and layers of creative criteria before making a decision. The system could surface what appeared strong, flag what deserved a closer look and bring the relevant context forward at the moment it was needed most.

The model was not used to make decisions. It simply helped a curators make better judgement calls with more of the organization’s accumulated learning in the room.

Putting things into practice

When it comes to scaling quality across an organization, the goal is never to automate taste. It is to make good judgment easier to exercise and explain under pressure.

That distinction matters even more as AI systems improve. Large language models can inherit criteria remarkably well. Give them a thoughtful rubric, useful context, strong examples and clearly bounded reasoning, and they can become capable partners in evaluative work.

What they cannot conveniently inherit is the creative formation that produced the criteria to begin with: years of looking, making, traveling, reading, comparing, failing, feeling, noticing nuance and changing your mind.

In the world of experiences, someone still had to notice that place should enrich an activity rather than simply host it. Someone had to understand why host credibility does not automatically produce a compelling run of show. Someone had to recognize that consistency and sameness are not the same thing.

AI can help us scale creative reasoning, but it cannot replace the formation of our judgment. Likewise, the infrastructure we build around creative judgment can never really be finished. Once other people begin putting standards and criteria into practice, reality starts talking back.

Maybe curators keep disagreeing about the importance of the same signal. Maybe a certain class of experiences repeatedly falls outside of acceptable criteria. Maybe guests love something you barely accounted for. Maybe a principle that worked beautifully in one market becomes absurd when applied somewhere else.

These moments of friction are not an inconvenience to the system, but new signals that inform the evolution of taste. When pieced together, they form an iterative operating system:

SIGNAL → TRANSLATION → INFRASTRUCTURE → SIGNAL

Defined standards help people notice quality. Their decisions produce new evidence. The evidence changes what the organization sees. Better seeing produces better language. Better language produces better tools. Over time, the operating system for taste evolves through encounters with reality.

Taste has to survive beyond you

In a scaled organization or marketplace, this is ultimately the responsibility of creative leadership: designing the conditions through which recognizable taste can travel, adapt and evolve with the market.

A good creative leader can make the right call. A great creative leader builds the language, rituals and systems that help other people understand why it was the right call — while giving them enough ownership to develop sound judgment in their own domains of expertise.

That doesn't mean the infrastructure of taste should eliminate disagreement. Quite the opposite. If everyone always reaches exactly the same conclusion, you may have built a compliance machine rather than a culture of creative judgment. The goal is to create enough shared understanding that disagreement becomes useful: I understand the standard, and I think this case asks us to interpret it differently.

This also makes quality more durable, because eventually you will not be in the room. One day there won’t be time for a leadership review. The creative director will be working on something new. The curator who developed the original instinct will move on. The organization will enter a new market. An AI system will encounter an edge case its prompt does not neatly resolve. And someone new will still have to decide what good looks like.

That is the real challenge of scaling taste — not turning subjective judgment into a formula or cloning one person's preferences across an organization, but creating enough shared language, examples and practice that people become more discerning without becoming less of themselves.

Which, in a strange way, brings me back to those walks with my mom. Nobody handed us a rubric for front doors. There was no neighborhood quality framework. We learned by paying attention together. By pointing things out. By disagreeing. By explaining ourselves. By seeing enough examples that our instincts became sharper.

Our shared language came from the looking. And eventually, the language changed how we looked.

Maybe that is the most important part of building an operating system for taste. The point isn't simply to make your judgment legible to other people, but to create the conditions for collective judgment to emerge, improve and evolve by learning what good looks like in practice.

Plenty of people have taste. Fewer can explain it. Far fewer can build a system that helps good taste survive beyond them.

The real test is not whether you'll know it when you see it.

Will anyone else?