9 March 20266 min read
Small teams tend to outsource in the wrong order. The instinct is to hand over whatever is most annoying, which is usually the thing with the most context attached, and context is exactly what an outside team does not have yet.
A better first candidate has three properties: it is well understood, it is low-context, and it is currently being done badly because nobody has time to do it properly.
Good first candidates
- The test suite everyone agrees is needed and nobody has written
- A mobile app mirroring a web product whose behaviour is already settled
- Integrations with third-party services, where the contract is a published API
- Reporting and data pipelines, where 'correct' is checkable against a source
- Accessibility and performance work, which is specifiable and measurable
- Support coverage in hours your own team does not work
What these share is that success is legible from outside. Someone who does not know your company's history can still tell whether the job was done.
Poor first candidates
- The core algorithm or model that differentiates the product
- Anything where the requirements live only in one founder's head
- Work that needs constant negotiation with a specific customer
- A rescue of a system nobody currently understands
The last one deserves a note. Handing a mess to an outside team is tempting precisely because nobody internally wants it, but an outside team will spend the first month discovering what the mess is, and you will pay for that discovery without gaining the understanding. Map it internally first, then outsource the fix.
Sequence it deliberately
Start with one clearly bounded piece rather than a broad engagement. You are testing two things at once: whether the provider is good, and whether your own team has the management capacity to direct outside people well. The second is failed more often than anyone expects, and it is much cheaper to discover on one workstream than on four.
Then expand along what worked
If the first piece goes well, expand into adjacent work where the team has already built context, not into something entirely different. Context is the expensive part, and reusing it is where outsourcing starts genuinely paying for itself rather than merely covering its own overhead.