CONSULT CIRCLE | CLOUD OPERATIONS
What you actually buy with a managed service, what you give up, and how to compare the two honestly.
The comparison is usually framed as cost against control, which is too simple. A managed service is not just someone else doing the same work. It changes what work exists, who is accountable when it fails, and what your team spends its time on.
The right answer differs by workload and by organisation, and it changes as teams grow or shrink. This is a framework for making the comparison properly rather than by instinct.
What you are actually comparing
| Dimension | Self-managed | Managed service |
|---|---|---|
| Direct cost | Lower unit price for infrastructure | Higher unit price, bundling operational work |
| Staff cost | Requires in-house skills, on-call cover and holiday cover | Absorbed by the provider |
| Control | Full control of version, configuration and tuning | Constrained to what the platform exposes |
| Speed of change | You decide when to patch and upgrade | Provider decides, within a maintenance window |
| Accountability | Yours entirely | Shared, with a service level commitment |
| Risk concentration | Spread across your team | Concentrated in a supplier relationship |
| Scaling operations | Effort grows with estate size | Largely absorbed as the estate grows |
| Exit | You own everything | Migration effort to change provider |
Table 1 — The real dimensions of the comparison.
The costs people leave out
Omitted from self-managed models
- Out-of-hours cover, including the cost of carrying on-call and the attrition it causes.
- Holiday, sickness and departure cover for specialist skills held by one or two people.
- Recruitment and onboarding when those people leave, which for scarce skills can take months.
- Training to keep skills current on a platform that changes continuously.
- The opportunity cost of senior engineers doing routine maintenance rather than higher-value work.
Omitted from managed service models
- Work that remains yours regardless, typically application-level ownership, data and access decisions.
- The internal effort of managing the supplier: reporting, escalation, service reviews.
- Change control friction where the provider controls maintenance windows and you do not.
- Exit cost, which is real and is rarely modelled at the point of signing.
A decision framework
Score each workload or platform against these questions.
- Is this a differentiator for your business, or is it plumbing? Differentiators are worth keeping in-house; plumbing usually is not.
- Do you have the skills today, held by more than one person?
- What is the out-of-hours requirement, and how is it covered now?
- How much control do you genuinely need over version and configuration, as opposed to how much you are used to having?
- What happens if the person who understands this leaves next month?
- Is the operational load growing with the estate, or is it flat?
- What would exit look like, and how long would it take?
Where each model tends to win
| Situation | Usually better | Why |
|---|---|---|
| Small team, broad estate | Managed | Operational load exceeds capacity and skills cannot be deep everywhere |
| Deep in-house expertise already funded | Self-managed | The cost is already incurred and control has real value |
| Strict change control or unusual configuration needs | Self-managed | Managed platforms constrain what you can change and when |
| 24x7 availability requirement, no existing on-call | Managed | Building on-call cover from scratch is expensive and hard to sustain |
| Rapid growth | Managed | Operational effort scales with the estate; managed absorbs much of it |
| Highly regulated with bespoke controls | Mixed | Often managed infrastructure with retained control of data and access |
Table 2 — Where each model tends to be the better fit.
Where to go next
- Cloud cost optimization playbook
- IT Procurement Services, for running a managed services selection properly
- Monitoring as a Service
Frequently Asked Questions
Is managed cloud more expensive than self-managed?
On unit price, yes. On total cost it depends whether you already employ the skills. If delivering it yourself means hiring, covering on-call and carrying key-person risk, the comparison is usually much closer than the headline rate suggests.
What do we lose control of with a managed service?
Typically version and patch timing, some configuration depth, and the ability to make changes outside a maintenance window. Whether that matters depends on the workload; for most it does not.
Can we mix both models?
Yes, and most organisations do. A common pattern is managed infrastructure and platform services with applications retained in-house, so the team focuses on what is specific to the business.
How do we avoid lock-in with a managed provider?
Agree exit terms at the outset, keep documentation and configuration in your own repositories, and make sure your team retains enough understanding to take it back. Exit cost is real and is best negotiated before signing.
What should a managed service actually include?
Defined scope and service levels, monitoring and response, patching and lifecycle, reporting, and a named escalation route. Anything not written down will be assumed to be yours when something breaks.
Talk to Consult Circle
We deliver managed monitoring and infrastructure services, and we will tell you honestly where self-managed is the better answer for your team. Book a free 30-minute call - 0203 916 5593 - info@consultcircle.com