It depends on frequency
If you will have a steady stream of automation and data work, developing internal capability pays. If you have one project, learning on it is the most expensive way to do it.
The worst outcome is someone in the business learning Python on a system the business depends on, without review, and leaving eighteen months later.
Where internal capability works
- Someone already comfortable with scripting or spreadsheets at an advanced level
- A steady stream of small automation work
- Time genuinely allocated, not squeezed into evenings
- External review of anything that matters
- Realistic scope — small automations, not core systems
A sensible progression
- Start with small, low-consequence automations
- Have someone external review the code
- Establish version control and testing habits early
- Move to more significant work as confidence builds
- Keep anything business-critical with support
What to keep external
| Keep external | Why |
|---|---|
| Anything the business depends on | Continuity risk |
| Security-sensitive work | Specialist knowledge |
| Anything with regulatory exposure | Consequence |
| The first production service | Patterns worth getting right |
| Machine learning | Genuinely specialist |
Support the person doing it
Someone learning alone, without review, will develop habits that are difficult to unlearn and will produce code nobody else can maintain.
Periodic external review — even a few hours a month — changes the outcome substantially and is inexpensive.