Cut your company into segments

The org chart is a map of who reports to whom. It tells you almost nothing about how work moves, which is why it's useless for finding clogs. A pipe diagram tells you how work moves and says nothing about reporting lines, which is why I draw one for every company I'm involved in.

Here's the exercise. It takes an afternoon and it is uncomfortable in a productive way.

Step 1: Follow one unit of work end to end

Pick something real that your business does repeatedly. A customer signing up. An invoice getting paid. A song getting released. Not the idealized version from the process doc — the actual version, the one that happened last Tuesday.

Now trace it. Every hand it passed through, every system it entered, every point where it sat and waited. Write each one down in order. Don't group, don't clean it up, don't skip the step where someone copied a value from one tool into another. That step is the most important thing on the page.

Most people expect eight steps and find twenty-three.

Step 2: For each step, fill in four columns

In — what must exist for this step to start?
Out — what exists when it's done that didn't before?
Owner — which role, not which person, is accountable?
Gauge — what number would tell you this step is healthy, and where do you read it today?

Fill these in honestly. The honesty is the whole exercise; a tidy chart of what should be true is worth nothing.

You will find three kinds of rows, and each one means something specific.

Complete rows are segments. Defined in, defined out, a role that owns it, a number you can actually read. These are the parts of your business that can be scaled, delegated, automated, or replaced without an act of archaeology. Protect them.

Rows with a blurry In or Out are situations. Work arrives here by custom rather than by definition — someone notices, or gets pinged, or checks a shared inbox. These are your clog candidates. They can't be handed off because there's nothing to hand.

Rows with a person's name in the Owner column are single points of failure. Not because that person is bad — usually the opposite; they're your best person, which is exactly how they ended up holding it. But a step owned by a name rather than a role is a step that leaves when they do.

Step 3: Circle every empty Gauge cell

This is the part that stings.

In most companies I've run this on, somewhere between half and three-quarters of the Gauge column comes back empty, or comes back with an answer like "we'd know if it was broken." That's not a gauge. That's a hope with a job title.

An empty gauge cell means that step is dark. Not necessarily broken — dark. It might be running beautifully. You don't know, and neither does anyone else, and the first news you'll get is the news that arrives after a customer notices.

Every empty cell is a to-do item. It doesn't need to be a dashboard. It needs to be a number, produced automatically, that a named role looks at on a stated cadence. "Median hours from signup to first successful import, posted in #ops every Monday" is a complete gauge. It cost an afternoon to build and it will pay for itself the first time it moves.

Step 4: Find the seam that hurts

Look at where work waits between rows. Not inside a step — between them. Handoffs are where companies leak, and the diagram makes them visible for the first time.

Ask two questions at each seam:

Does the receiving step have everything it needs the moment work arrives? If the answer involves anyone going back to ask a question, the upstream Out and the downstream In don't match. That mismatch is a fitting that doesn't fit, and it will cost you a little bit of every single unit of work forever.
Would anyone notice if work stopped arriving here? If a queue can quietly go empty — or quietly grow — with no alarm, that seam is dark.

Fix the fittings before you touch anything else. It's cheaper than reorganizing, it doesn't require anyone to change jobs, and it compounds: every unit of work through that seam gets faster from then on.

What you do with the diagram

Three things, in order.

Instrument the dark steps. Cheapest, fastest, highest leverage. You cannot prioritize what you cannot see, so this comes before any restructuring.
Convert situations into segments. Write the In and Out explicitly. Publish them where the people doing the work can see them. This is not documentation for its own sake — the definition is what makes the step replaceable, and replaceable is the goal.
Widen the constraint. Only now. With gauges in place and interfaces defined, the actual bottleneck is visible, and it's frequently not the one everybody assumed.

Why this is worth an afternoon

Because from here on, growth is assembly.

When your business is a list of segments with defined fittings and live gauges, adding capacity is a decision rather than a project. You know which segment is at its limit, because the gauge says so. You know what a replacement has to accept and produce, because the In and Out are written down. You can hand a segment to a new hire, a vendor, or an agent, and evaluate the result against the same number you were already watching.

That's what "everything connects" buys you. Not elegance — optionality.