- Measure a real workflow before choosing a tool.
- Separate setup, operation and maintenance.
- Keep capacity value distinct from cash savings.
Full cost has four lines
The initial quote covers scoping, building, testing, documentation and go-live. Then add licences or API calls, hosting, monitoring, fixes when a format changes and human time to review exceptions.
A cheap build may be expensive to run. A more robust integration may cost more upfront and require fewer interventions later. Compare options over the same period, such as twelve months.
| Cost | Questions to ask |
|---|---|
| Setup | Scoping, access, development, tests, training |
| Operation | Licences, APIs, compute, hosting, storage |
| Maintenance | Formats, APIs, models, security, support |
| Review | Exceptions, human approval, failure recovery |
Measure the workflow, not a hunch
For a representative week, count cases, active minutes per case, rework and exceptions. Record the starting input and expected output. This baseline lets you test a prototype on the same cases.
Microsoft describes process mining as a way to see how processes actually run and find bottlenecks. For a small business, a simple log is often enough: volume, duration, exception rate and owner.
- Real monthly volume, including seasonal peaks
- Active work time rather than total waiting time
- Exception rate and recovery time
- Loaded labour cost or internal value of the time involved
Time released is not automatically money saved
If automation releases 58 hours a month, you first gain capacity: more cases, shorter delays or time for useful work. It becomes a cash saving only when an expense actually disappears, such as overtime or outsourced work.