Automating a Business That Is Changing Every Month

Growth changes what breaks

A process that works at twenty units a day fails at two hundred, and the failure is rarely where anyone expects. Usually it is a handoff between people rather than the work itself.

Model your current process at three times the volume before deciding what to build. The bottleneck you find is the thing to automate, not the thing that annoys you today.

Automate what scales linearly

• Intake — every new order or enquiry is another keying job
• Customer communication, which grows faster than volume because busier operations generate more exceptions
• Invoicing and payment chasing, where a backlog becomes a cash flow problem
• Scheduling and allocation, where the combinatorics get worse with scale

Leave the parts you are redesigning

If you are still arguing about how a process should work, encoding one version makes the argument more expensive. Growth businesses argue about a lot of processes, and that is healthy.

Pick the two or three that have been stable through the last doubling. Those will survive the next one.

Build configurable, deliberately

1. New product types, sites and categories added without a developer
2. Thresholds and routing changed by your team
3. Roles and permissions extensible rather than hard-coded to today's org chart
4. Volume assumptions checked — a batch that works at 200 records may time out at 2,000

Expect to revisit within a year

That is not failure; it is what growth does. Budget for a review and a phase two rather than treating the first build as final.

The alternative — waiting until the business settles — usually means waiting through the period when the admin burden is doing the most damage.

Frequently asked questions