Organisations that run online appointment desks, record systems and messaging portals eventually face an operational question that has nothing to do with medicine: who watches the systems for misuse, and how? SOC Ops, which is about keeping a monitoring team dependable day and night, offers three broad answers. You can build a team, buy one as a managed service, or mix the two. This article compares them on the dimensions that matter to administrators, using evidence summarised by the Ops Atlas. It is general operational information for non-clinical readers and not advice on any specific system or regulation.
The yardsticks used here
The Atlas, which is open about the fact that the label “SOC Ops” is its own convenient tag for a long-standing function, frames the work as turning alerts into timely, recorded decisions. Its page on the Atlas description of SOC Ops lists scope items that make a sensible comparison grid: staffing around the clock, tiers, playbooks, handover, metrics, and the choice between building and sourcing. We will judge each model on four of them: coverage, accountability, quality of decisions and learning.
Option A: build in-house
Coverage. The arithmetic is unforgiving. MITRE’s estimate, as cited by the Atlas, is that one seat staffed at every hour needs around five people, and a minimum of two analysts should be on shift at any moment. For a hospital-sized administration or a small telehealth operator, that is a large payroll for a function that is invisible when it works. Many in-house teams therefore cover office hours only, leaving nights to chance.
Accountability. Clear. The people who watch the systems work for the organisation that owns them.
Decision quality. Potentially high, because analysts learn local context: which departments travel, which vendors send odd traffic, which integrations are noisy. That context turns many alerts from puzzling to ordinary.
Learning. Strong, provided someone writes things down. In-house teams can run weekly reviews, adjust procedures and repeat a yearly self-assessment using the open SOC-CMM model.
Option B: buy a managed service
Coverage. The main selling point. A provider spreads the five-people-per-seat cost across many customers, so overnight and weekend cover becomes affordable.
Accountability. Here the Atlas is direct, citing MITRE: owners and executives remain accountable, whatever the contract says. A provider takes on the watching; it cannot take on your risk appetite. Administrators should therefore insist on named escalation contacts, a written test for when an event becomes an incident, and a defined list of actions the provider may take unprompted.
Decision quality. Variable. Remote analysts lack local context, so more alerts arrive as “undetermined” and more questions come back to you. Contracts that measure the provider on how fast tickets close make this worse, since the Atlas lists time-to-close incentives among the anti-patterns that raise reopen rates and rush investigations.
Learning. Depends on reporting. Ask for medians and 90th-percentile figures by shift and severity, because a single mean can hide a bad night.
Option C: hybrid
A common compromise has an internal owner for the queue during working hours and a provider for nights and weekends. The Atlas does not prescribe it, saying plainly that no single design suits everyone, but the arrangement tends to score well on the grid. Coverage is affordable, local context survives, and accountability is straightforward. The price is a handover between two teams, which is where things get lost.
That makes the pass-down record essential. The Atlas describes a shared log that both the leaving and arriving teams initial, in which anything not yet accepted stays with whoever held it before. Administrators who already run staffing handovers at a front desk will recognise the principle.
Side-by-side summary
- Cost shape: build is mostly fixed payroll; buy is a recurring fee; hybrid sits between.
- After-hours cover: build is hard; buy is easy; hybrid is easy if the handover is disciplined.
- Local context: build is strongest; buy is weakest.
- Accountability: stays with the organisation in every case.
- Typical weakness: build struggles with nights; buy struggles with context; hybrid struggles with the seam between teams.
A short thought experiment
Picture a mid-sized appointment platform that handles a few thousand sign-ins a day. An unusual login pattern appears on a Sunday at midnight. In an in-house model with office-hours cover, nobody sees it until Monday. In a purely managed model, a remote analyst sees it at once but may not know that the organisation’s clinics run a regional campaign that explains the traffic, so a question comes back and time passes. In a hybrid model, the provider checks the shared notes, sees the campaign entry, closes the alert as benign with a reason, and the Monday owner reads the log. The scenario is generic and invented, yet it shows that the quality of shared notes often matters more than the choice of model.
Tools matter less than sequencing
Whatever model you choose, the sequencing advice in the Atlas is the same. Gather and search logs first; automate later. The reference passes on a caution from the Australian signals agency that workflow tooling seldom fits young programmes. When automation does arrive, the Atlas recommends narrow guardrails: automated isolation reserved for rules that have earned trust, limited to everyday staff equipment and throttled to a set pace. Critical systems remain a human decision, and every silenced rule needs a named keeper and a date on which it lapses.
Questions worth asking any option
- How is each alert closed, and is a reason recorded?
- What is the median and 90th-percentile wait before a person first looks, split by shift?
- Whose name is on the rota in the small hours of a public holiday?
- Which written procedures exist for the incident types you see monthly?
- How is maturity assessed, and how often?
Related reading on this site
Money questions run through every choice above, and our walkthrough on splitting shared bills through a recurring tracker with clear rules shows the same logic of fixed versus variable costs. For readers curious about the digital side of consultations, there is also a plain-language overview of how online ENT care is introduced to the people using it; it is a reminder of how much sensitive information passes through everyday platforms.
Bottom line
No option removes responsibility, and none works without records. The strongest comparison point is honesty about the quiet hours: ask each model to show you its worst night and judge from there.