We talk about how we work at Sanctuary more than most companies probably do. We like process.
What we didn’t do, for a long stretch of years, was document much of it in writing. (Typing, in this case.)
We’d settle something in a meeting. How a handoff should work. What a client is owed in the first week. Who owns a decision and who doesn’t. Why we picked one goal over the other one. These were real decisions, made by people paying attention. Then they went into the air, and we moved on to whatever was next.
Everyone heard the words.
The trouble is that people heard the words differently. Or heard them the way they needed to hear them. Or forgot them by March.
We had a shared belief that we had a process or a decision. A shared belief that we were aligned. The problem is that a shared belief isn’t something you can see or point at. It’s a group of people who each remember things a little differently and haven’t found out yet.
Saying It Out Loud Isn’t Documenting It
We’ve talked about the “rule of seven” in meetings over the years, the one that says people need to hear or see something seven times before it sinks in. I looked it up again recently. It comes from movie studios in the 1930s trying to work out how many times someone had to walk past a poster before they’d buy a ticket. There’s no study underneath it, though it has held up well enough in my experience.
There is better evidence for the underlying idea. Two researchers, Tsedal Neeley at Harvard and Paul Leonardi at Northwestern, followed thirteen managers across six companies for more than 250 hours and recorded every message those managers sent and received. About one in seven communications turned out to be deliberately redundant: the same request sent a second time through a different channel. The managers who did that moved their projects along faster and with less friction than the ones who said it once and assumed it landed.
So repetition works. It just has a ceiling, and the ceiling is that spoken messages keep changing shape. Every person who hears it filters it through their own experience, their own workload, and their own needs. Months later they’re following the version they remember, which is usually a version nobody ever said out loud.
You can say it in a Monday meeting. You can say it again in the quarterly. You can say it a third time, one-on-one, with feeling.
And a year later you’ll watch it come back done in a way you never described, by someone who is certain they heard it right.
And when two people remember it differently, there’s nothing to check. You’re just comparing memories.
The Ladder from Checkboxes to Vision
The most useful way I’ve found to think about this is as a ladder, from the smallest thing you do to the largest thing you’re trying to become.
At the bottom is a checkbox on a task. Did the file get named correctly? Did the second reviewer actually look at it? Boring, unglamorous, and the place where consistency is either won or lost.
Above that sits the process itself, and this is the rung most people mean when they say documentation. The best book I know on the topic is Process! by Mike Paton and Lisa González. Their argument is that a company of nearly any size has somewhere between six and ten core processes. People, Marketing, Sales, Operations, Accounting, Customer Retention. Name them, get the leadership team to agree on what they’re called, and give each one a single owner. Then, most importantly, document at 20/80, capturing the twenty percent of steps that drive eighty percent of the result. A handful of the most important steps with a few sub-bullets. Two pages, not twenty. Gino Wickman compresses the same idea into one chapter of Traction and lands on the phrase we’ve used ever since: documented and followed by all.
Above the process are the people. Roles belong on paper, not in an assumption. In How to Be a Great Boss, Wickman and René Boer put “keep expectations clear” first among the management practices, and they mean it literally. The accountability chart with the handful of roles under each seat. The quarterly priorities. The numbers each person owns. Written, not implied. Almost every performance conversation I’ve had that went badly was one where two people had different unwritten ideas about what the job was.
Their three-strike system is a good example. It only works because each strike is a conversation you document afterward, with specifics and a deadline. Most managers hate that part, or they feel their words are enough. But if you ever reach the third strike, you have a record, and more importantly, the person across the table has never once been surprised.
Above the people are the goals, the values, and finally the long-term vision, written in language plain enough that somebody in their second week can read it and know where the company is going.
What I think tends to happen is that companies document the middle of that ladder and skip both ends. They’ll build a decent process manual but never write down a single checklist underneath it. Then they’ll wonder why the work still comes out inconsistent. Founders will talk extensively about the goals and the vision. Then they’ll wonder why their team has completely different interpretations of where the company is headed.
Both ends matter. Documenting the checkbox is how the work stays precise. Documenting the vision is how a hundred small decisions point in the same direction when nobody’s in the room to ask.
Documented and Followed Until It’s Time to Improve
Documentation without adoption is theater. You get the satisfaction of having written it down and none of the benefits. So you train the document, you manage the document, and when someone finds a better way to do step three, you change the document that week. A stale document is worse than no document, because now the team has decided the written version is fiction and gone back to working however they personally feel is best.
Toyota has been running on a version of this for sixty years. The written description of the current best way to do a job is called standardized work, and it’s treated as a baseline rather than a final answer. Kaizen, their word for improvement, means small changes made by everyone, continuously. When someone on the line finds a better way, the improvement isn’t finished until the written standard has been changed to match. The document is what makes the improvement real. Everything before that is one person doing it differently or better than everybody else and nobody knowing about it or knowing why.
It’s the opposite of what people expect documentation to do. Writing it down doesn’t freeze the process. It’s the thing that makes the process possible to improve, because now there’s a version everyone can see.
The same worry shows up as a creative one. I own a marketing agency, so I’ve heard it plenty: all of this will flatten the work. Paton and González answer it better than I can: freedom within a framework. The documented part is the scaffolding that shouldn’t require a decision, written down precisely so people can spend their judgement on the work that actually needs judgement. Nobody’s best thinking has ever gone into remembering which folder the assets live in.
What Growth Exposes
When you’re small, everything living in a few heads might be fine, because those heads are in the room. Then you hire, and hire again, and the person who knew how it all worked is on vacation, or busy, or gone. That’s when you find out whether you built a company or a group of talented people with good memories.
Companies without documentation don’t usually fail dramatically. They just fall behind. Every departure costs a month. Every new hire spends a quarter reverse-engineering the job by watching whoever is nearest, which means they inherit that person’s habits along with the work. Then they teach it to the next person the same way.
My lesson is that anything that only lives in conversation isn’t a process, or a goal, or a vision. It’s a rumor. Write it down, keep it short, make it specific, put a date on it, and give people something they can read and point to instead of something they have to decipher and remember. A company that can’t point at its own words is running on memory, and memory doesn’t scale.