Your Systems are Situational

Your systems are situational. Hear me out.

My personal systems were always two systems: one for work, one for home. And they worked incredibly well.

At work, I existed on Slack reminders and Gmail's snooze feature. The system was essentially this: things would pop up when they needed my attention, and not before. It worked for years. At home, I used Apple Reminders. It was not clean, but it had geofencing and time-based alerts, so I could say "remind me when I get home" and get a nice little pop-up at the front door.

The cracks showed one day when I was on an airplane, flying from Warsaw back to Amsterdam.

Around that time I had started developing ideas for a few side projects (including my advisory business, Otterstein.org #shameless plug). And the side projects didn't cleanly fit into either system. They weren't work, so I couldn't use Slack to remind myself. I didn't have Gmail set up for the side business yet because it was still in idea-mode and I was on an airplane without wifi. And my personal system didn't fit either, because the projects needed more management than a simple reminder was capable of.

So somewhere over Germany I had to admit it: both systems had to come down. Not patched. Torn down, and something completely new built in their place. Something durable that accommodated not just my personal life, not just my professional life, but a potentially infinite number of future passion projects (And if you know me, you'll know that when I like to solve problems, I like to solve a single problem that solves a 1000 later problems).

Rebuilding, I held myself to three rules. The capture mechanism had to be as frictionless as possible and identical for all three territories. The reminder mechanism had to be located in a single interface. And all the three had to stay clearly delineated from each other. I built the workflows with the least negative friction I could manage. I'm a strong believer in the power of positive constraints and positive friction, but this had to fit simply into how I already move through a day.

And what this became for me was a realization: the systems I had made were not durable. They were situational. They worked for the situation they existed in. But as soon as I evolved, and this is not a company speaking (this is a very personal one), I needed a system that worked for the new version of me.

We don't know what the future versions of us will need. But we can build systems with durable components. If you assume you will have more interests in the future — and you will — then you can assume you'll need to capture those things, to be reminded of those things, to manage the tasks that come with those things. The interests are unknowable. The needs are not.

The systems we build need to account not for an unknowable future, but for a very likely one.

Companies do the exact same thing, at scale. The approval flow that works at ten people, the priorities that live in one head — these are situational systems wearing the costume of durable ones, and they hold right up until the company evolves into a new version of itself. That's the pattern I wrote about in my previous little article Everything breaks at 3 and 10, but it took being stuck on a plane ride from Warsaw alone with my thoughts for me to notice I'd been doing it personally.

You Can Measure Trust With a Stopwatch

An executive I audited answered the question with a number and nothing else. The question: how long does it take for bad news to reach you once something breaks. Her whole answer: two days. No story, no complaint. The tone of someone reporting the weather, and the tone is the finding. A two-day delay on the information she most needs fresh had become a fact of her company, filed somewhere near the payroll date.

A few questions later, the same intake asks what her team doesn't tell her. Her answer: that they disagree with leadership's decisions, and that they don't feel supported.

Read those together. She knows the silence exists. She can name its shape. What she can't do is make it move faster. We dug into everything she'd tried. Always two days.

I now ask every company for this number, because it behaves consistently: bad news travels at roughly two days per level. One layer of management between the problem and the decision-maker, two days. Two layers, four. The exact figure varies; the compounding doesn't. Good news arrives the same afternoon. It carries itself, because people volunteer whatever makes them look good.

The extreme case shows the mechanism. A small restaurant team had their card system die mid-shift. The server told nobody. He took cash where he could, waved a few tables through, got the evening done. The owner found out at close, counting a drawer that made no sense. "The card system is down" was worth everything at seven and almost nothing at midnight.

Colin Powell: bad news isn't wine, it doesn't improve with age. The kitchen version is blunter. Problems age like milk.

So why two days per level? Bad news pays a toll at every level it climbs, and the toll has three booths.

Verification first. Nobody wants to raise a false alarm, so whoever saw the problem checks it again, asks a colleague, waits to see whether it happens twice. A day gone.

Softening second. The message gets translated into safer words on the way up. "The client is furious" becomes "some concerns on the client side." "This will slip three weeks" becomes "the timeline is tight." Each translation buys the messenger comfort and costs the listener meaning. By the time the words are safe enough to say, they're too vague to act on.

Timing third. The carrier waits for the right moment, or tries to fix the thing quietly and report a solved problem instead of an open one. Solved reads as competence; open feels like exposure. Sometimes the quiet fix works and you never learn how close it came. Sometimes it fails, and the problem arrives late and grown.

Notice what the booths are made of. Verifying is diligence. Softening is tact. Fixing before reporting is ownership. The delay is built out of virtues, which is why there's no villain to find, and why announcing "I want to hear problems earlier" changes nothing. You'd be asking people to stop being careful, considerate, and self-reliant. They won't, and they shouldn't have to.

Here's the reframe that earns its keep. We treat trust as a feeling and measure it, when we measure it at all, with a survey. Treat it as a speed instead. Trust is the transmission speed of true information, and the bad-news lag is its gauge: when did the thing break, and when did the person who could act on it hear? That gap is your number. You can measure trust with a stopwatch.

Ron Westrum built a whole typology of organizations on this axis: what a culture does with inconvenient information, and above all what it does to the people who carry it. Which is why "my door is always open" doesn't move the number. An open door only helps the person already standing in it.

What moves it is a rule about messengers. Report first, fix second, for anything touching delivery or a client. Raw wording is welcome; stale is the only failure. Then the half most companies skip: every messenger gets thanked, visibly, with the message judged separately from the messenger's performance. One punished carrier closes the road for everyone who watched it happen.

Leadership complies first. Norms are set by what leadership visibly does. The founder who opens Monday with her own freshest bad news, the deal she misread, the deadline she blew, does more for transmission speed than any anonymous survey. She's demonstrating, at her own expense, what happens to people who say true things: nothing. They get heard.

One condition, and it's load-bearing. This assumes your last messenger walked away whole. If the most recent person who carried bad news upward got a lecture about attitude, the channel is dead before it opens, and you owe a debt before you can ask for anything new: credit, publicly and by name, someone who brought you a hard message. Pay the old toll before announcing the road is free.

You'll know within a few weeks: at least one piece of bad news reaches you less than a day old, worded plainly enough to sting. If everything arriving is still stale and pre-cushioned, the booths are still collecting, whatever you announced.

The lag never goes to zero, and it doesn't need to. A team that checks before escalating is a team worth having. Know your number, then watch it the way you watch cash. Two days can become one. One can become an hour. Every hour you recover, the problem you hear about is younger, smaller, still an afternoon's work instead of a client call.

Unlike the survey, the stopwatch can't flatter you.

Nobody ever kills anything

There's a question I ask on my diagnostic intake that does more work than it suggests: what have you deliberately stopped doing in the last year?

The answer that comes back most often isn't a thing. It's a standard. "We stopped accepting mediocre work, even on the small stuff." It's said with pride, and the pride is absolutely earned. The company meant it, and the work did get better.

Read it as an operations answer rather than a culture answer, though, and it inverts. Raising a standard retires nothing. It adds. Every piece of work now has a higher bar to clear, which can mean another review pass, more rework, more senior time spent on things that used to ship at good-enough. It means that someone has to become the quality police. The company took on load and filed it under "stopped." That's the expensive trick in it: a standards raise pays out the emotional credit of an abandonment while delivering none of the relief. You feel lighter and your calendar gets heavier.

Peter Drucker named the thing they were reaching for and missed: planned abandonment. His test was blunt. For everything you currently do, ask, "if we weren't already doing this, would we start it today?" Where the answer is no, stop. Not trim, not deprioritize, not do-it-quarterly-instead. Take it off the list and tell people it's gone.

Companies skip this for a structural reason, not a character one. Adding is a decision one person can make in a moment, alone, and it resembles leadership. Stopping needs a reason, a defense, and someone willing to own it if the thing turns out to have actually mattered. So the list of what a company does moves in one direction only, while the people it lands on stay the same people. Nobody decided to overload them. Everybody decided to add one more good idea.

Here's what I find most telling. Companies that never plan an abandonment still abandon plenty. They just don't get to choose which. The choosing gets done under load, by the calendar, around four o'clock on a bad week.

One companyI work with wrote out its full recurring meeting list. The daily project scrums survive. The weekly status check-ins survive. The weekly management development session, the monthly per-department curation meeting where the next stretch of people's work would be planned... and half the 1:1s get pushed, then pushed again, then quietly stop appearing. The person who wrote the list flagged those three as the most valuable meetings on it.

That is an abandonment decision. Nobody made it. What survives under load is the meeting where another person is waiting on you, because cancelling it costs you an awkward conversation today. What dies is the meeting whose payoff is six months out and whose only stakeholder is one person's growth. Urgency wins because urgency is loud.

And it feeds itself. Development sessions are precisely what would make a team need less supervision. Cancel them and the team stays dependent. The dependence gets paid for in check-ins, course corrections, and questions that only leadership can answer, and those hours come out of the next development slot, which is the easiest thing on the calendar to move. Each turn of that loop tightens the next one. Name the loop and you can predict it.

So two little tweaks, both cheap, both about actively choosing the abandonment instead of receiving it.

The first is a standing abandonment hour, once a quarter, with a date on it in advance. List everything the company does on a recurring basis: meetings, reports, rituals, service lines, tools, the newsletter nobody reads. Ask Drucker's question of each one. The rule that makes the hour real is that it doesn't end until one genuine recurring commitment is retired, publicly, with a name attached to the decision. One is enough. Not a standards raise, not "we'll do less of that." An actual retirement, announced as a win, so that stopping something reads as competence rather than surrender. This assumes the company is genuinely over capacity; if it isn't, the hour becomes theater, and the tell is simple, nobody can say what the retirement freed up.

The second is to invert the cancellation rights on one meeting. Take the most valuable recurring session that keeps getting cancelled, and make it cancellable only by the person it exists to serve, never by the person running it. Move whatever status content can be written down into writing, so the surviving meetings stop being the only place information travels. This assumes those meetings die from lack of time rather than lack of substance. If they're being cancelled because nobody quite knows what to do in them, that's a content problem, and defending the slot will only produce a protected, awkward half hour.

The sign to watch for is near-term. Withina few weeks you should be able to point at one thing that is no longer on the list, and at one calendar slot with an unbroken attendance record. If the quarterly hour produces only standards raises and reshuffled priorities, it's being run by the same instinct it was designed to interrupt, and you'll know because the list will be exactly as long as it was in the spring.

The instinct behind "we stopped accepting mediocre work" is a good instinct. Someone was trying to say something true about the company becoming more serious about its craft. It just isn't a stop, and calling it one costs you the thing you actually needed, which is a shorter list.

Everything on that list was added by someone who thought it was a good idea, and most of them were right at the time. The question isn't whether it was a good idea then. It's what would happen if it quietly went away. And the next time the calendar makes that choice on your behalf, look at which meeting it picked. Its taste is remarkably consistent, and it isn't yours.

Idle and overloaded at the same time

There's a company just past ten people where the leadership team works systemic overtime, and where, on any given afternoon, someone is sitting idle, waiting to be told what to do next.

Both things are true at once. The founder wrote them down three questions apart on a diagnostic without noticing the collision: "not enough people for how much we want to do," and, a page earlier, teams that need constant guidance and check-ins to make plans and stay on track.

Hold those two sentences next to each other. If the company of that size were actually short on people, nobody would be idle. Everyone would be saturated with clear work and delivery would still slip, purely on volume. That's not what's happening. The work exists, the people exist, and the thing connecting them, a decision about what to do next, is stuck behind an invisible wall.

Eliyahu Goldratt built a whole discipline, the Theory of Constraints, around one observation: every system has exactly one constraint at a time, and the throughput of the whole system is set by it. Adding capacity anywhere except the constraint changes nothing. His examples were factories, but the logic doesn't care what the machines are made of. You find the constraint the same way on a shop floor or in a tech company: find the clog. Where does work pile up waiting?

In this company, and in most companies that describe themselves this way, the clog forms in front of leadership. People aren't waiting for materials, budget, or tools. They're waiting to be convened. Waiting for the plan to be blessed, the priority to be confirmed, the next piece of work to be assigned. The constraint isn't headcount. It's leadership attention.

Here's why the contradiction survives so long without anyone naming it: waiting wears a costume. Nobody sits with folded hands until the next decision arrives. They fill the gap with something, inbox, tidying, a third polish pass on work that was already fine. The waiting gets silently absorbed into the appearance of busyness, so it never shows up anywhere. Leadership's overload, meanwhile, is extremely visible: they're the ones answering messages at midnight. The company sees one half of the picture and writes its story from it. Everyone is maxed out. We need to hire.

That story is the expensive part, because hiring is one mistake that makes this worse. A new hire arrives needing the one resource the company has least of. They need direction, context, decisions, check-ins, all drawn from the same pool of leadership attention that was already the bottleneck. You've added demand to the constraint and booked it as capacity. Six months later the overtime is worse and the payroll is bigger. The prediction is uncomfortable but reliable: when leadership attention is the constraint, headcount growth slows the company down before it speeds it up.

The counter-evidence usually sits in plain sight, and it did here too. When a decision didn't have to wait, the company moved fast. A minor internal policy came up one morning, was decided the same day, and rolled out within three. Nothing about this company is slow. Decisions it makes are quick and clean. Decisions waiting to be made are where all the time goes. That distinction matters, because it means the fix isn't "get better at deciding." The deciding works. The access to it doesn't.

Goldratt's answer to a constraint you can't instantly remove: exploit it, then subordinate everything else to it. Translated out of factory language, stop spending leadership attention by the drip, on whoever happens to catch someone in a hallway or a chat thread, and start spending it in deliberate, scheduled doses. Two mechanisms that I recommend do mooooost of the heavy lifting.

The first is a greenlight session: one recurring, bounded meeting where the queued decisions actually get made, and every decision leaves the room with a name and a date attached. Not a status meeting. Status can move to writing. This session exists only to convert waiting into direction, and its rule is that nothing leaves undecided; the allowed outputs are yes, no, or not-yet with a specific revisit date. Name the owner and you can predict the follow-through.

The second is gating your intake. If everything that gets started must pass through the session, the system gains something it never had: a place where "not yet" can be said by a rule instead of a person. Most small companies start work the moment it's imagined because no mechanism exists to hold it. The gate is that mechanism, and it protects the constraint from the thing that floods it fastest, which is new work nobody finished deciding about.

This, of course, assumes people are waiting on decisions. If they're waiting because they don't know how to do the work, rather than which work to do, you have a capability gap, and no amount of decision hygiene will fix it. The two look similar from a distance. The test is cheap: for one week, whenever someone is between things, write down what exactly they were waiting for. If the list is full of "confirmation," "approval," "the meeting," it's attention. If it's full of "help," it's skills. And if it's attention, you should see the line visibly shorten within two to four weeks of the first greenlight sessions; if it doesn't, the diagnosis was wrong, and the week of notes will tell you where to look instead.

This leader's instinct, "not enough people," wasn't foolish. It's the natural reading when everyone looks busy and everything is late. It just aims the money at the wrong spot. Before you write the job posting, spend the week watching what people are actually waiting on. Companies are usually not short of hands. They're short of identified work to put in them, and that's not a hiring problem. That's the operating system begging for an update.

If you have to chase it, you don’t have commitment infrastructure

A manager I worked with kept a mental list of things people had promised. Not a spreadsheet, not a board. A list in her head, and every few days she'd work through it: did you send that, is that done, where are we on the other thing.

She thought the problem was the people on the list. Some of them were reliable and got asked anyway, which stung: they'd done the work, and now they were being checked on like they hadn't. Some of them weren't reliable, and being asked was the only reason anything happened at all. Over time the unreliable ones learned they didn't need to remember their own commitments. She'd remember for them. Why hold two copies of the same information in your head?

That split matters more than it looks. Ask a reliable person the same question twice and you've told them, without meaning to, that their word wasn't good enough the first time. Ask an unreliable person the same question twice and you've confirmed that nothing bad happens if they wait for the ask. The chasing trained both groups, just in opposite directions, and neither direction was the one she wanted.

Here's the tell: her own team's task board was empty. Not because nothing was happening, but because nothing that happened had ever been written down anywhere except her memory. The commitments existed. The infrastructure to hold them didn't.

Chris Argyris had a name for the gap this creates: the difference between what an organization says it values and what it actually rewards, moment to moment. She said she valued trust and autonomy. What she rewarded, in practice, was proximity to her attention. The people she chased hardest were the ones she trusted least, and everyone on the team could see the ranking even though nobody had said it out loud.

Chasing feels like management. It produces motion: a message sent, a status pulled, a box that can be mentally checked. But chasing is where leadership's attention goes to die. Every minute spent asking "did you do the thing" is a minute not spent on the ten things only she could do. And it doesn't scale. The list in her head could hold maybe fifteen open items before things started falling off it silently, which is worse than never asking, because now the drop looks intentional.

The team read the chasing as distrust. It wasn't, exactly. It was the absence of anywhere else for a commitment to live. When there's no visible place for "I said I'd do this by Thursday," the only enforcement mechanism left is a person asking another person, repeatedly, and that always reads as surveillance even when the intent is care.

The fix isn't a lecture about ownership. It's a place. A visible board, physical or digital, doesn't matter which, where commitments get written down when they're made, not when someone asks about them. And the board has to start with leadership's own items, publicly, before anyone else's items go on it. Ask a team to make their promises visible while yours stay in your head, and you've just built a surveillance tool with your name off it.

This is the part most managers skip, because it's uncomfortable in a way process changes usually aren't. Putting your own open items on a public board means the gap between what you promised and what you delivered is now visible too, to the same people who used to only see yours enforced on them. Most of the resistance to visible-commitment systems isn't the team resisting. It's the leader realizing the board cuts both ways.

Once commitments are visible, chasing changes shape. Most of it disappears, because a glance at the board answers the question that used to take a message. What's left is real: something that's overdue with no explanation on the board is now data, not a hunch. That's a conversation worth having, and it's a completely different conversation than "hey, did you get to that" for the ninth time.

The manager rebuilt her team's board that quarter and put her own five open items on it first. Two things happened. The reliable people stopped feeling watched, because the board answered for them before she had to ask. And the unreliable pattern got easier to see and name, because for the first time it wasn't buried inside her memory. It was sitting on a board where everyone, including her, could see who kept their word and who needed the system to remember for them.

None of this required new software or a framework with a name. It required one thing to exist that hadn't existed before: somewhere other than a manager's head for a promise to live.

If you're maintaining a mental list of who owes you what, that list is the whole problem. It's not a discipline issue on their side. It's an infrastructure gap on yours.

---

Not sure whether this is your pattern? The fit-check takes two minutes, no email required.

Everything Breaks at 3 and 10

There's a restaurant I work with that grew from three people to five and fell apart.

Nothing dramatic happened. Nobody quit, nobody stole, the food didn't change. The owner just noticed, over about two months, that everyone was exhausted, quality was slipping, and the team that used to feel like a kitchen felt like a queue. Everything routed through him. It always had. That was the point: at three people, the owner is the operating system, and it works beautifully.

At five, it doesn't. And the cruel part is that nobody changed anything, which is exactly the problem. The system that made the place good stopped fitting the place it had become.

Hiroshi Mikitani, who built Rakuten, put a number on this: everything in a company breaks every time headcount roughly triples. At 3, at 10, at 30, at 100. Not the people. The invisible stuff: who decides what, how information moves, what gets rewarded, what a meeting is for. The coordination you got for free at one size becomes the thing quietly taxing everyone at the next.

The tell is always the same sentence: "it used to just work."

It did. Osmosis worked. Everyone heard everything because everyone was in the room. Approval was a look across the pass. Priorities didn't need writing down because there was only one, and everyone was staring at it.

Then you tripled, and osmosis died, and nobody held a funeral. So people fill the gap the only ways they can: they wait, or they silently absorb. When work is quietly absorbed it sneaks up on you. It's always just a little outside of your scope at a time until your scope becomes something unrecognizable. You're a few "can you justs" away from a total mental and physical collapse.

When teams wait for the founder, the founder absorbs every decision to keep things moving, and "nothing is stalled" becomes technically true while everything gets slower. The founder reads their own exhaustion as personal inadequacy. It almost never is. It's an operating model built for three, running at ten.

Two things worth doing the week you notice this.

Name the threshold out loud. "We tripled, so our old ways expired" is a diagnosis nobody has to be ashamed of, and shame is the main reason founders fix the wrong thing. Name the problem and you can predict it.

Then rebuild the two or three mechanisms the old size gave you free. Who can say yes to what, written on one page. Where information lives when it's not in your head (which pays exponential dividends). What the team would say the priorities are, without checking.

The restaurant did this. The owner kept her morning market run and insisted on a "family shift meal", wrote the routine decisions onto one page and handed them to her second, and got back the hours that made him good in the first place.

If some version of "it used to just work" has been rattling around your head: it's not you. It's probably the system needing an update.

---

Not sure whether you're at a threshold? The fit-check takes two minutes, no email required. — linked as above.