Growth is arriving faster than the operation can absorb it

The three things that break, and how to tell them apart

Capacity

Queues grow steadily, everyone is busy, quality holds until it suddenly does not. Test: does throughput rise proportionally when you add a person for two weeks?

Process

Work is re-done, clarified and chased. Test: how much of the week is spent finding out what state something is in rather than advancing it?

Ownership

Work stalls with nobody accountable, and escalations converge on one or two people. Test: pick five stalled items and ask who owns each — count how long the answer takes.

Stabilise first, then scale

1. Cap the intake temporarily where you can

   Slowing acquisition for a few weeks is cheaper than a churn cohort and a burnt-out delivery team. This is a commercial decision, so it needs to be made explicitly rather than by exhaustion.

2. Find the constraint step

   One step governs throughput. Measure wait time before each step, not touch time inside it — the queue in front of the constraint is where the delay actually lives.

3. Protect the constraint

   Remove non-essential work from it, make sure it is never idle waiting for inputs, and stop measuring the utilisation of everything else as if it mattered equally.

4. Fix ownership at the boundaries around it

   Handoff contracts on both sides — see work leaks between teams.

5. Add capacity only where it moves throughput

   Adding people anywhere except the constraint increases coordination cost and nothing else.

6. Re-open the tap deliberately

   Increase volume in steps, with a defined signal that says stop. Growth without a stop condition is how the same fire restarts next quarter.

What growth exposes, in order

Volume does not create new weaknesses; it makes existing ones expensive at a predictable sequence of thresholds.

First: the informal handoff

Work that moved because two people sat near each other stops moving when a third team is added. This breaks before capacity does, and it is usually misread as a resourcing problem.

Second: the single expert

The person everything routes through becomes the constraint. Their queue grows, their quality drops, and the business discovers how much undocumented process lived in one head.

Third: measurement

Reporting that worked at twenty units a week stops being readable at two hundred. Totals hide the exceptions, and leadership loses the ability to see where the loss is.

Fourth: standards

Under load, the standard becomes whatever fits in the time available. Quality variance rises before any metric records it, and customers notice first.

The signals worth watching afterwards

• Wait time in front of the constraint step, weekly.
• Exception volume by boundary — rising exceptions predict the next break.
• Rework rate: how often a unit of work returns to a previous stage.
• Time to first value for new customers, which degrades before churn shows up.

Related work

The audit sequence is in the operations audit playbook. If acquisition is the thing outpacing delivery, the unit-economics view in digital growth is the other half of the conversation.