Mapping a business as a system: the walkthrough

You map a business as a system in six steps: name the purpose, draw the boundary, list the stocks, trace the flows, mark the loops and delays, then find the constraint. It takes one afternoon, a whiteboard, and a tolerance for discovering that the business in your head is not the business on the wall. The map is usually wrong in one specific, expensive place — which is the point of drawing it.

This is the practical companion to my systems-thinking guide for founders. That piece explains the concepts; this one is the pen-in-hand procedure.

What are you actually drawing?

Three kinds of object, borrowed from Donella Meadows. Stocks — accumulations you could count on a given day: enquiries waiting, proposals out, active clients, cash, your own remaining attention. Flows — the rates that fill and drain them: outreach per week, proposals sent, projects completed, invoices paid. Loops — the places where a stock feeds back to change a flow: more clients produce more referrals, which produce more enquiries.

Not org charts, not process swim-lanes. Those show who reports to whom and who clicks what. A system map shows why the numbers move.

What are the six steps?

The mechanism, in order — each step gates the next:

1. Name the purpose. One sentence, judged by behaviour rather than the website. When the diary says the system exists to keep the founder busy, then that is its purpose, whatever the mission statement claims.
2. Draw the boundary. Decide what is inside (things you can change) and what is weather (market rates, buyer moods). Ten minutes, but done honestly it kills half your excuses.
3. List the stocks. Aim for six to ten boxes. If you cannot state today's number for a box, that is a finding — you are running a stock you cannot see.
4. Trace the flows. Arrows in and out of every stock, with real weekly rates where you know them. When a stock has an outflow but no reliable inflow — enquiries drain into proposals but nothing refills enquiries — then you have found a future famine, months before the P&L shows it.
5. Mark the loops and delays. Circle the reinforcing loops (compounding) and balancing loops (capping), then write the delay on each arrow. Delays are where founders misjudge: a proposal sent today is revenue in ninety days, so panic-selling in a bad month fixes a quarter that is already lost.
6. Find the constraint. One flow governs the throughput of the whole map — Goldratt's bottleneck. Ask of each candidate: if this doubled tomorrow, would revenue follow? Only one honest yes survives.

What does a real map look like?

A worked fragment from our own wall. Stocks: prospect database → active sequences → positive replies → booked calls → proposals → clients. Flows between them, with rates. The revealing part was the leftmost arrow: every downstream stock depended on the database inflow, and that inflow was a person doing manual list work — slow, error-prone, and skippable in busy weeks.

The map made the constraint undeniable, so we automated it; the build log of the AI agent that builds and enriches a B2B database documents exactly what that took. Without the map we would have "improved sales" by rewriting copy — polishing a flow that was never the constraint.

What will the map reveal that you didn't expect?

Three findings recur across nearly every firm I have drawn this with. An invisible stock — usually goodwill or founder energy — that everything depends on and nothing measures. A missing inflow — one stock, often top-of-funnel, refilled by luck rather than a flow anyone owns. And a misplaced fix — the current improvement project sitting nowhere near the constraint.

Expect an argument, too. When two partners map the same business differently, the difference is the real conversation, and it is cheaper to have it over a whiteboard than over a bad year. Choosing the mental models that earn their keep beforehand shortens that argument considerably.

What do you do with it on Monday morning?

Three things. Put a number and an owner on every stock nobody could quote. Schedule one intervention at the constraint — only there. And date the map: redraw it quarterly, because the constraint moves once you relieve it.

What you should not do is treat the map as the whole truth. Some of what matters most — morale, reputation, the way the team behaves when you are not in the room — appears on no arrow, because culture is a system output, not a box you can draw. The map earns its keep anyway: it tells you where to push, and it stops you pushing everywhere at once.

---

Keep the thinking coming: The Format — a fortnightly letter on business and personal systems. Subscribe by email.