5
min read
The broken windows theory of brand
What NYC's subway cleanup can teach us about brand drift
Sofya Leonova
Co-founder + Marketing Director
One of my favorite theories about human behavior is the Broken Windows Theory, and recently I've realized that this theory, and the experiments behind it that were conducted in New York in the 1980s, have a lot to teach us about brand consistency.
The theory in a nutshell
Social scientists James Q. Wilson and George Kelling introduced the broken windows theory in 1982. When a broken window in a building goes unrepaired, people walking past it draw a conclusion that nobody is paying attention here, nobody is minding the place. That conclusion is an invitation. Another window gets broken, then another, and the same logic spreads to bigger things, until a neglected block becomes a place where serious crime feels normal. Their argument was that visible, unaddressed disorder lowers the bar for the next person, and that small neglect, left alone, compounds into serious crime.
How the theory applies to brand
The analogy holds for brand, though not because brand is about aesthetics. A broken window doesn't invite crime because broken glass is dangerous. It invites crime because it broadcasts that nobody's watching, and an off-brand sales deck or a rogue landing page does the same thing. The real damage is downstream: the artifact tells the next person that the rules aren't enforced here, so their shortcut is fine too, and the cost of the next bit of disorder drops a little. One unfixed thing becomes an invitation to the next.
The analogy is useful, but it stops fitting a company’s brand in three specific places, and those three places are where the practical lessons live.
The person breaking the window is your best employee
In a neighborhood, the person breaking the window is a vandal. In your company, the person shipping the off-brand deck is your strongest AE, who needed a one-pager by Friday and built one because the real assets were three Slack threads and a permission request away. The demand gen hire who grabbed a default font wasn't defying the brand, she was routing around friction. Your people are good employees optimizing locally, and they'll keep taking the fast path for as long as the fast path and the on-brand path are two different paths. The vandal model tells you to catch and punish, but there's nothing here to catch. There's only a system that made the wrong thing easier than the right one.
The enforcer becomes the bottleneck
The second place the analogy breaks is one the theory created for itself. Broken windows got operationalized as zero-tolerance policing in 1990s New York, and one of the most-cited parts of its legacy is the backlash: enforcement that came down hard on small things bred resentment, fell heaviest on the people with the least power, and never touched the conditions creating the disorder. The same thing happens with brand. A brand team that exists purely to gatekeep becomes a bottleneck, and bottlenecks get routed around. Your AE still needs the deck by Friday, only now he builds it without telling anyone, because telling anyone means a wait and a lecture. The unraveling doesn't stop. It just moves out of sight, hidden from the team that believes it finally has control.
So you do need a champion. What that champion can't be is someone who only enforces, because the job is to lower the barrier so the windows stop breaking, rather than to patrol them. A real champion writes simpler policies, removes the approval steps that don't earn their cost, and puts the right templates and assets where people already work, so the on-brand move is also the fastest one.
AI has the potential to be either the fastest and easiest way to unravel a brand or the tool that lowers the barrier to creating on-brand work quickly and seamlessly. One of the ways we address this is by building an AI-native BrandOS — a set of markdown files that codify both the verbal and visual brand — as part of every brand project. Unlike a Figma brand guidelines file that no one will ever reference, it enables internal teams to produce on-brand work where they already work: increasingly inside AI tools. The goal is to make staying on-brand the path of least resistance, rather than an extra tax that a busy team will quietly opt out of paying.
A brand has to keep evolving
The third place the analogy fails is the biggest one, because neighborhoods don't need to be redesigned to stay relevant and brands do. A perfectly enforced, frozen brand is its own kind of failure: order with no life in it. The thing you're protecting also has to grow, which means a real brand champion is doing two jobs at once, holding the line against entropy and evolving the system so it still fits the company a year from now.
Why Series A is the riskiest time to get this wrong
Your brand is new and thin. You have very few windows, so each broken one is proportionally enormous. And you're likely hiring faster than at any other point in the company's life, so a flood of new hands touch the brand every month with no memory of why any of it is the way it is. The conditions for compounding disorder are never more present, and there's usually no one whose job it is to notice the first crack. The answer is to lay down the easy path before the hiring accelerates, while there's still little enough to organize that doing so is cheap.
And if you're already past the clean-slate moment, the theory carries its reassuring half too. When New York went after the disorder, it started with the subway cars: scrub the graffiti off every one, and refuse to put a tagged car back into service until it was clean again. In the decade that followed, the city recorded one of the steepest crime declines an American city had ever seen. Restoring order to the small, visible things is what pulled the bigger behavior back into line. The broken window wasn't incidental to the crime. It was the permission slip.
Brand drift works the same way. The off-brand deck and the rogue page are the permission slip for the larger disorder: the messaging that wanders, the positioning that softens, the five salespeople who each describe the company a little differently. Scrub the small surfaces and you don't just get a more consistent brand, you revoke the permission, and curb the bigger drift. And it's never too late to start. Scrub the graffiti off and you reset the clock. The only penalty for waiting is volume, because the longer you let it drift, the more there is to scrub before you're back to a surface worth keeping clean.
