Why Growth Stops Paying: The Coordination Cost Nobody Books | HL Hunt
Why Growth Stops Paying: The Coordination Cost Nobody Books
Growth is discussed as one thing getting bigger. It is two things moving at different rates: output, which rises with headcount, and coordination, which rises with the relationships between people. Six staff to twelve doubles output and quadruples the communication surface. On a stylized firm that is 8.2% of revenue at twelve people and triples by twenty-four, almost none of it on any schedule anyone reviews. That gap is why the firm that doubled twice does not feel twice as profitable.
What you'll learn
The core thesis
Growth is discussed as though it is one thing getting bigger. It is two things moving at different rates: output, which rises roughly with headcount, and coordination, which rises with the number of relationships between people.
Those are not the same curve. Adding a person adds one unit of output and adds a connection to everyone already there.
| People | Output index | Communication pairs |
|---|---|---|
| 3 | 3 | 3 |
| 6 | 6 | 15 |
| 12 | 12 | 66 |
| 24 | 24 | 276 |
| 40 | 40 | 780 |
Going from six people to twelve doubles output and quadruples the communication surface. Nothing was mismanaged. The arithmetic of pairs is just different from the arithmetic of people.
Our position is that the cost of coordination is the main constraint on small firm growth, that it is almost never on anyone's books, and that the stages where businesses stall are the points where it overtakes the output it was supporting.
Putting a number on it
Take a firm with $1,450,000 of revenue and 12 staff, which is about $120,833 per head. Suppose coordination, meaning meetings, status updates, handoffs, clarifications and rework caused by missed information, consumes 4.5 hours a week per person at a loaded cost of $46 an hour.
- Annual coordination cost: $119,232
- As a share of revenue: 8.2%
Now let the hours rise with size, as they do, and the shape of the problem appears:
That is the whole mechanism, and it explains a pattern owners describe constantly without a name for it. Per our expansion analysis, revenue grows, headcount grows, and profit does not, which gets attributed to pricing or to a bad hire. Per our capacity analysis, the capacity that disappeared went into coordinating the capacity that remained.
Per our owner compensation analysis, it is also why the owner's own hours are the first thing consumed. The person who holds the most context is the one every clarification routes through, so coordination load lands disproportionately on whoever is least able to delegate it.
Where firms stall, and why there
The stall points are not arbitrary. They are where a coordination mechanism that worked stops working.
| Size | What coordinates the work | What breaks |
|---|---|---|
| 1 to 3 | Everyone knows everything | Nothing; this is the efficient size |
| 4 to 8 | The owner holds all context | The owner becomes the bottleneck for every decision |
| 9 to 15 | Informal specialization | Nobody holds the whole picture and nothing records it, per our institutional memory analysis |
| 16 to 30 | Process, or chaos | Process has to be written, which nobody has time to do |
| 30 plus | Management layer | Overhead that produces no output and is therefore cut first |
The 9 to 15 band is where most small firms get stuck and it is the hardest transition. Below it, the owner can hold everything. Above it, written process is unavoidable. Inside it, the firm is too large for one person's head and too small to afford the overhead of writing things down, which is why it is a band rather than a line.
The last row describes why firms that make the transition often reverse it. A management layer is pure coordination cost: it produces no billable output and its contribution is the absence of problems, which is per our institutional memory analysis exactly the kind of value that cannot be evidenced and is therefore removed in the first difficult quarter.
A falsifiable claim
Revenue per head should peak at a small size, fall through a middle band, and recover only after a firm has written process and added a management layer. That is testable on any dataset with firm size and revenue: plot revenue per employee against headcount within an industry. If it is flat or rises monotonically, coordination cost is not a binding constraint and this article is wrong. We expect a dip, and the location of the dip should be the band where firms in that industry actually stall. A finding that the dip occurs at different sizes in different industries would be consistent with the mechanism rather than against it, because the work determines how much coordination each handoff requires.
What makes scale worth it anyway
Firms do grow, and the gains are real. Any account of coordination cost that implies everyone should stay small is wrong.
- Fixed costs spread. Per our fixed cost analysis, compliance, software, insurance and premises cost roughly the same for eight people as for eighteen.
- Specialization raises output per hour by more than coordination takes back, up to a point. That point is the question, not whether the gain exists.
- Concentration risk falls. Per our concentration analysis, a larger firm survives losing a customer that would have ended a smaller one.
- Capacity wins work. Per our capacity analysis, orders a small firm must decline go to someone, and that someone becomes the incumbent.
- Credit access improves. Per our business credit analysis, a larger firm with a longer record borrows on better terms, which compounds.
The second bullet is the honest centre of the question. Specialization and coordination pull in opposite directions, and the optimum is wherever they cross for a particular kind of work. A firm doing varied bespoke work crosses early; one doing repetitive standardized work crosses much later, which is why the same headcount is comfortable in one trade and chaotic in another.
Reducing the coordination load
The useful move is not staying small. It is lowering how much coordination each unit of output requires.
- Reduce handoffs, not meetings. A meeting is a symptom; each handoff between people is a place where context must be transferred and can be lost.
- Make a decision owner for each class of decision. Most coordination is people seeking permission, and per our exception analysis an undefined decision right routes everything to whoever is most senior.
- Write down the reason, not the procedure. Per our institutional memory analysis, a one-line rationale per rule is cheap and prevents the questions that generate most clarification traffic.
- Standardize the common case and route the rest. Per our legibility analysis, coordination cost is concentrated in the unusual jobs, so separating them protects the standard flow.
- Measure something each person can act on alone. A metric requiring three people to interpret generates meetings about the metric.
- Buy coordination where you can. Scheduling, invoicing and payroll systems are coordination mechanisms sold as software, and per our build versus buy analysis building them is rarely the right call at this size.
- Add the management layer before it is unavoidable, and accept that its output is invisible, because adding it during a crisis is adding overhead at the worst moment.
Item one is the highest-return change and the least intuitive. Firms attack coordination cost by cutting meetings, which removes the mechanism and not the need, so the information moves to interruptions instead. Reducing the number of people a piece of work passes through lowers the requirement rather than the symptom.
Scenarios and what we are watching
| Scenario | What we would expect |
|---|---|
| Coordination tooling keeps improving | The stall band moves upward, and firms pass through sizes that used to stop them without noticing |
| Work becomes more bespoke | The crossover arrives earlier, and the efficient firm size in those trades falls |
| Remote and distributed work spreads | Coordination cost per handoff rises, which lowers the efficient size unless process compensates |
| Nothing changes | Firms continue to stall in the 9 to 15 band and continue to attribute it to hiring or to the market |
The third row is the one we would flag as underappreciated. Distributed work removes commuting and premises cost, which is measured, and raises the cost of each context transfer, which is not. Per our reference point analysis a firm has no benchmark for what its coordination load should be, so a rise in it registers as everyone being busy rather than as a cost that grew, which is precisely the condition under which it keeps growing.
Frequently asked questions
Output rises with headcount; coordination rises with the pairs between people. Six to twelve staff doubles output and quadruples the communication surface.
Usually between nine and fifteen people: too large for one person to hold the context, too small to afford writing the process down.
No. Fixed costs spread, concentration risk falls, and capacity wins work. The question is where specialization gains stop exceeding coordination costs.
Reduce handoffs, not meetings. Cutting meetings removes the mechanism and not the need, so the traffic becomes interruptions instead.
Key takeaways
- Output rises with headcount and coordination rises with communication pairs, so six to twelve staff quadruples the coordination surface.
- On a stylized firm, coordination cost is 8.2% of revenue at twelve staff and rises from $39,744 to $397,440 between six and twenty-four.
- Headcount quadrupling while coordination cost rises tenfold is why revenue growth and profit growth come apart.
- The 9 to 15 band is the hardest transition: too large for one person’s head, too small to afford writing process down.
- A management layer produces no billable output, so its value cannot be evidenced and it is cut in the first difficult quarter.
- Specialization and coordination pull opposite ways, and the crossover depends on how bespoke the work is.
- Reduce handoffs rather than meetings: cutting meetings removes the mechanism, not the need.
- Distributed work lowers measured costs and raises the cost of each context transfer, which registers as everyone being busy.
This analysis is educational and does not constitute business, financial, or management advice. All figures are stylized illustrations using assumed revenues, headcounts, hourly costs, and coordination hours, and do not describe any particular firm. The relationship between headcount and coordination load varies substantially by industry, work type, and operating model, and the communication-pairs calculation is an upper bound on relationships rather than a measure of actual information flow.