Managed services vs hiring a support team: which is cheaper at your size

01The coverage arithmetic nobody runs

Genuine 24x7 coverage with a sustainable rotation needs at least four to five people. Fewer than that and you are running an on-call rota that burns people out, produces attrition, and eventually fails at 3am because the one person who understood the system left.

This is why small organisations buying internal 24x7 support tend to end up with something that is 24x7 on paper and best-effort in practice. If you genuinely need overnight response, the honest internal cost is four to five salaries, and comparing that to a retainer changes the answer substantially.

02Where the crossover sits

For business-hours support of a small estate, an internal hire is often cheaper and better - context accumulates, response is immediate, and one competent person can hold it. The external option struggles to beat that on cost at small scale because a vendor still has to staff, onboard and manage the relationship.

For 24x7, or for an estate diverse enough to need several specialisms, external usually wins on cost because the vendor amortises a rota and a skills mix across multiple clients. You are buying a share of a team rather than funding a whole one.

03The hidden cost of internal support

Support work interrupts. When your product engineers carry support, the cost is not their time on tickets - it is the context switching that makes their feature work slower and worse. Teams routinely under-count this because the hours are invisible; the delivery slowdown shows up as an unexplained drop in velocity.

There is also a career problem. Engineers who spend most of their week on support tend to leave, and the ones most capable of fixing root causes leave first. That turnover cost is real and rarely appears in the comparison spreadsheet.

• Context switching costs more than the ticket hours.
• Support-heavy roles have higher attrition.
• The best engineers leave those roles first.
• Velocity loss is invisible in the support budget.

04The hidden cost of outsourcing

The main one is knowledge dispersal. A vendor accumulates operational knowledge about your system, and if the relationship ends that knowledge can leave with them. This is manageable - require runbooks, documentation and monitoring configuration as contractual deliverables - but it must be required, because it does not happen by default.

The second is distance from the product. An external team that only handles incidents will not spot the design decisions creating them. That is why the improvement allocation matters: it converts a support relationship into one where somebody is paid to reduce the incidents rather than absorb them.

05The arrangement most mid-market teams land on

One internal engineer who owns the relationship, understands the estate and handles business-hours escalation, with an external team providing overnight and weekend coverage plus surge capacity. You get accumulated context and genuine coverage without funding five salaries.

This works when the boundary is explicit. Agree who owns which decisions, what the external team can change without asking, and how escalation flows in both directions. Ambiguity there produces incidents where everyone assumed the other party was handling it.

• One internal owner for context and escalation.
• External for out-of-hours and surge capacity.
• Explicit decision boundaries in both directions.
• Improvement allocation contracted, not assumed.