The Org Chart Is Fine. The Workflow Is the Problem.

A founder told me last month she'd hired three operations managers in two years and the same orders still shipped late. The boxes on her org chart kept moving. The late shipments didn't. The problem was never who owned the work. It was the route the work had to travel to get done.

The reflex everyone reaches for first

When something keeps going wrong, almost everyone reaches for the same tool. The org chart. Add a manager. Move someone senior over the problem. Reshuffle the team and hope the new arrangement holds.

And the thing keeps going wrong.

I see this in nearly every business I look at closely. The people are good. The structure is reasonable. Nobody is lazy and nobody is hiding. Yet orders ship late, refunds take eleven days, the same client gets onboarded twice. So the founder concludes it must be a people problem, because people are the only thing the org chart lets you adjust.

That's the trap. The org chart is a map of who reports to whom. It says nothing about how work actually crosses the company. And the work crossing the company is where almost all the friction lives. When your only tool is the chart, every problem starts to look like a hire, a firing, or a title change.

I want to make a quiet case for looking somewhere else first. Not because reorganizing is always wrong, but because it's usually the second move dressed up as the first.

The thing nobody actually designed

What's usually broken isn't the structure. It's the path the work takes to get done. Who hands off to whom. Where it waits. How many times the same approval happens twice. That path is the workflow, and here's the strange part: in most companies, nobody designed it.

It grew. Someone left and someone filled in. An email chain became a habit. A habit became "how we do it here." A temporary workaround from a busy quarter quietly turned permanent. No one sat in a room and decided the refund should pass through four inboxes. It accumulated, the way paper accumulates on a desk you keep meaning to clear.

So you end up with a route that no single person chose and no single person can see end to end. The salesperson knows their stretch of it. Finance knows theirs. Support knows theirs. Each piece looks fine from inside. The handoffs between them — the seams — are where things fall through, and the seams belong to no one.

This is why "who dropped the ball" is almost always the wrong question. The ball didn't get dropped. It got handed to a place where nobody was standing, because nobody ever decided who should stand there.

Why the path beats the person every time

W. Edwards Deming spent his career teaching organizations how systems actually produce results. He put it about as plainly as it can be put: a bad system will beat a good person every time. That line is worth sitting with.

It means you can hire around a person, but you cannot hire around a broken path. The path will win. Put your most capable employee inside a process with a hidden double-approval and a four-day wait, and the output will still be slow. Then you'll look at the slow output and quietly wonder if you hired the wrong person. You didn't. The system you handed them was built to produce that result, and it did.

There's a mechanical reason this holds. The speed and quality of any repeated process are set by its weakest link and its longest wait, not by the talent of the people touching it. A chain moves at the pace of its slowest junction. Adding a stronger person three steps away from the bottleneck changes nothing — the work still piles up at the same junction it always did.

This is also why heroics don't scale. When a great employee pushes work through a broken path by sheer effort, the path stays broken and invisible. The day they leave or burn out, the breakage reappears at full strength. You were never running a working system. You were running a person who was compensating for a broken one.

The exercise: follow one real thing end to end

Before you change who does the work, watch the work move. Here's the exercise, and it costs you an afternoon.

Pick one thing your business does over and over. An invoice. An onboarding. A refund request. Don't pick the dramatic edge case — pick something ordinary that happens every week. Now take one real example, a specific one with a name on it, and follow it from the very start to the very end.

Write down every stop. Every inbox it sits in. Every time it changes hands. Every approval. Every place it waits for a person to notice it. Be honest about the waiting, because the waiting is usually most of the elapsed time and almost none of the work.

You will find things. The step that exists because of one bad week three years ago, still running long after the reason expired. The approval nobody remembers requesting, signed off without reading. The form that gets re-typed into a second system by hand. The handoff that happens over a Slack message that isn't logged anywhere, so when that person is on holiday, the work simply stops and no one knows why.

Keep two numbers next to each step: how long the work itself takes, and how long it sits waiting to be picked up. The gap between those two columns is your real problem, and it has nothing to do with anyone's job title.

"But our work is too varied to map"

The most common pushback I hear: our work is creative, or custom, or different every time — you can't put it on a flowchart. Sometimes that's true. Often it's the broken path defending itself.

Most businesses are far more repetitive than they feel from the inside. The work feels varied because every instance has its own details and its own small fire. But the path is usually the same five or six moves: a request comes in, someone scopes it, someone approves it, someone does it, someone checks it, someone delivers it. The content changes. The route rarely does. If you genuinely can't describe how a common task travels, that's not proof the work is too complex to map. It's the clearest sign that nobody has, and that the variation you're feeling is chaos wearing the costume of craft.

The other honest caveat: mapping the path doesn't mean rigidly automating every step or stripping out human judgment. Some waits are real — a contract should be read, a diagnosis should be considered. The goal isn't to remove every pause. It's to tell the difference between a pause that adds something and a pause that's just a place the work goes to sit. Once you can see which is which, you keep the first and delete the second. You can't make that distinction until the whole path is in front of you on one page.

What changes when you can see the path

Once the path is visible, your options multiply. A reorg gives you exactly one lever: move a person. A map gives you a dozen. Delete the dead approval. Merge two steps that never needed to be separate. Move a check earlier so errors get caught before, not after, three more people have touched the work. Let one role own a seam that currently belongs to no one. Most of these cost nothing and require no new hire.

This is also where automation finally earns its place. Not as a buzzword bolted onto a process you don't understand, but as something you apply with precision once you can see exactly where the work waits and where it gets re-typed by hand. Automating a clear path removes drudgery. Automating a broken path just makes the breakage happen faster.

There's a quieter return, too. When the path is designed instead of inherited, the company stops depending on a few heroes holding it together by memory. New people get productive in weeks instead of months, because the route is written down instead of folklore. The founder gets to step out of the middle of every handoff.

So the next time something keeps going wrong, resist the chart for one afternoon. Pull one real example and follow it to the end. The fix is almost never a new box. It's a path someone finally looked at. If you want a structured way to map yours, that's the work we do at MARSA — marsa.ai/business.