A business outgrows its operating model long before anyone says so out loud. The signs arrive one at a time, each looking like a separate problem with a separate owner. Each gets a separate fix.
Here are the six that show up most often. Taken individually they look like bad luck or bad hires. Taken together they are the signature of a business running on a model built for a smaller version of itself.
1. Good people are firefighting. Adding more does not help
The first instinct when a team falls behind is to grow it. Sometimes that works. When it does not, the shape of the problem is usually different from the shape of the solution being applied. That reflex has its own post.
If output does not rise in proportion to headcount, the constraint is not capacity. It is the way work moves. New people inherit the same workarounds, need the same undocumented context and queue behind the same approvals.
2. Work gets done twice or falls between teams
Duplication and gaps are the same fault seen from opposite ends. Both mean ownership is unclear at a boundary.
Ask two managers who owns a step and you get two confident answers. Neither is lying. The business changed shape while nobody redrew the line.
This is why reorganising rarely fixes it on its own. A new chart moves the boxes without settling who decides what at the seams, which is where the friction actually lives. The common shape is a quote that delivery quietly rebuilds from the job sheet, because the sales version has been wrong before.
3. Customer experience slips while the support team grows
This one is counterintuitive enough that it usually gets misread as a people problem. Response times lengthen, complaints rise and review scores soften, all while the team handling them gets bigger.
What is normally happening is that support has become the place where upstream faults surface. The team is not underperforming. It is absorbing the cost of problems created elsewhere.
That absorption is worth measuring, because it is usually invisible. Every hour spent explaining a billing error caused upstream is an hour charged to support in the accounts while the cause sits in another team's process. Grow support and the symptom eases. The cause carries on producing work.
4. Refunds, errors and rework creep up
These rarely appear as a line anyone owns. They get absorbed into cost of sale and nobody revisits them, because each one individually looks like the cost of doing business.
The number worth knowing is what they total across a year. Then how much of that total is avoidable rather than inherent. Credit notes booked against the job they relate to are the usual reason nobody has that total: correct accounting that assembles the figure nowhere.
5. Nobody can get a straight answer on the numbers
The tell is not that reporting is missing. It is that reporting exists while people quietly work around it.
Once one person builds a side spreadsheet to check a figure, others follow. Before long each team runs on its own version of the truth and every meeting opens with a discussion about whose number is right.
6. Decisions stall or funnel through two people
Growth adds decisions faster than it adds decision-makers. If the same two names appear on every approval, the queue is invisible in the accounts but very real in the calendar.
Nothing fails outright. Things simply take longer. A capital request sits five weeks, not because anyone objects but because the two people who can approve it are in every other queue as well. The delay never shows up as a loss anywhere.
The second cost is quieter. People stop bringing options forward once they learn the answer always comes from the same place. A business can lose its best thinking this way without a single argument, because nobody experiences it as a decision being blocked. They simply stop asking.
Why they are worth reading together
Each of these has an obvious local fix:
- Hire into the stretched team
- Rewrite the process at the boundary
- Add reporting nobody argues with
- Escalate faster
Local fixes work when the problem is local. When six of them appear at once, the common cause sits underneath all six. Fixing them one at a time treats the symptoms in sequence while the cause stays put.
That is the case for looking across the whole operation before spending on any single fix. Not because the local fixes are wrong, but because the evidence decides which of them actually matters.
That is the case set out in where scaling businesses leak value: find where the money is going before committing spend to fix it.
Two things are usually true at once. Value is leaking through the gaps. Capability the business already owns is going unused, because nobody has the room to pick it up.
How to put a number on each of the six
Recognising a signal is the easy half. Sizing it is what turns an impression into a decision. Every one of these can be sized from systems you already run. Every one of them reduces to a calculation simple enough to sit on a dashboard.
| Signal | Where to look | The calculation | What it tells you |
|---|---|---|---|
| 1. Firefighting that hiring does not fix | Payroll and headcount in Xero or MYOB, against revenue by quarter | Revenue ÷ FTE = revenue per head | Flat or falling while headcount climbs means capacity was never the constraint |
| 2. Work done twice or dropped | Job or ticket system, filtered to items reopened or reassigned | Reopened jobs ÷ jobs closed = reopen rate | Ownership is unclear at a boundary. Split it by team to find which one |
| 3. Support growing while experience slips | Helpdesk tags in Zendesk or HubSpot, against order or case volume | Support contacts ÷ orders = contacts per order | Rising means upstream faults are surfacing downstream |
| 4. Refunds, errors and rework creeping up | Credit notes in the ledger, refunds in Stripe or Tyro | Credits and refunds ÷ revenue = leakage rate | What a year of small write-offs actually costs once assembled in one place |
| 5. No straight answer on the numbers | The files feeding your board pack, in Microsoft 365 or Google Workspace | Spreadsheets in the chain = hands per figure | How far a number travels before anyone is willing to trust it |
| 6. Decisions funnelling through two people | Approval history in your finance or project system | Median days request to decision = decision lag | Where the queue sits and whose name is on it |
Revenue per head is the one to put on a dashboard first. It is the cheapest of the six to calculate. It uses two figures you already close every month. It moves slowly enough that a twelve-month trend line reads as a trend rather than as noise. Contacts per order is the natural second, because between them the two answer the question underneath all six: is this business getting harder to run as it grows?
Each of these is a standing metric rather than a one-off exercise. Set them to calculate monthly and give each one a threshold that triggers an alert, so the trend arrives on its own instead of waiting for somebody to go looking.
The sixth is worth a service level as well as a metric. Most businesses have a published target for responding to a customer and none at all for responding to their own people. Something as plain as an approval older than five working days being named at the leadership meeting will change the number within a quarter.
Where your own figures will be wrong is worth knowing in advance. Reopen rate flatters any team that closes and re-raises a job rather than reopening it. Contacts per order understates the load if support takes phone calls that never become tickets. Leakage rate misses anything written off as a discount at the point of invoicing rather than credited afterwards. Treat the first read as a baseline rather than a measurement, then watch the direction it moves.
How many did you recognise?
That is the useful question, because one or two is normal in any growing business. Four or five is a pattern. A pattern means the cause sits underneath rather than inside any of them.
If you recognised four or more, the next question is which one is actually costing the most. Most leadership teams can name the signals confidently. Far fewer can rank them by what they are worth, because that ranking needs evidence nobody has assembled.
The Free Health Check scores the six areas these signals sit in, in about three minutes. If the picture looks about right, how we work explains what a full evidence-based read involves.