A team proposes self-hosting a managed service to save £80000 a year in licence fees. What do you add to their calculation?
Show the full answer Hide the answer
What is being tested
Whether you can complete a total cost of ownership calculation, and specifically whether you count the term that does not appear on any invoice.
What is missing
Run effort, which is usually the largest term. Patching, version upgrades, capacity management, backup verification, monitoring, security review, and — the one that dominates — on-call. A service that pages someone consumes far more than the hours spent responding, because it consumes the rotation's sustainability.
Half an engineer's ongoing attention is a substantial annual cost, and at that level of commitment the £80,000 saving has already evaporated.
Build effort. Time to first production use, including the work to reach parity with the managed service's availability, backup and failover behaviour — which is normally where the estimate is optimistic.
Opportunity cost. What those engineers would otherwise have built. For a small team this is the largest number in the calculation and the one never written down.
Infrastructure. The compute, storage and network the self-hosted version consumes — plus the redundancy required to match the managed service's availability, which is frequently omitted from the comparison because the managed service's redundancy was invisible.
Risk-adjusted cost. The expected cost of the outages that a less mature operation will have. A managed service has been operated by specialists for thousands of customers; your first year will not match it.
Exit cost, in both directions.
The questions to ask
- What is the run effort estimate, honestly? If the answer is "a few hours a month", ask who has operated this before and what their experience was.
- What availability does the managed service provide and what will yours provide? If lower, the difference has a business cost.
- What happens when the person who set it up leaves? A single owner is a real risk.
- Is there a specific capability the managed version does not offer? That is a legitimate reason; "cheaper" usually is not.
When self-hosting genuinely wins
- Scale where the premium becomes material. The threshold is much higher than teams assume, and it is a calculation.
- A required capability the managed version does not expose.
- Regulatory constraints the provider cannot satisfy.
- Existing deep expertise and a team whose job it is.
The strongest public example — a large storage workload moved off public cloud onto owned infrastructure — met all of these: one component, dominating the entire cost base, at enormous scale, with a team built to operate it. Those conditions are narrow, and citing that case as general support for repatriation proves considerably less than it appears to.
The framing to use
Convert both options to cost per unit of business value over three years, including people. That makes them comparable and connects the decision to something the business already measures — and it usually settles the argument without anyone having to be told they are wrong.