
IT roles: who does what
The titles overlap, shift from one organisation to the next, and on their own say little about what someone does. That is awkward when you have to decide who you are looking for. This is what sits behind the usual names.
There is no standard. The same title covers different work at an organisation of thirty than at one of three thousand, and that is not sloppiness but a consequence of scale: the smaller the team, the more hats per person.
The most useful distinction is not the title but the question of what someone maintains and what they build. Administration roles keep running what exists; building roles put down what does not exist yet. Hire a builder for maintenance work and you lose them within a year, and the other way round.
The question behind a vacancy is almost never "which title". It is: which work is not getting done, and how many hours a week is that. That answer decides whether you recruit, contract in, or outsource.
Why the titles shift
There is no body that fixes these job names. What a system administrator does is set by the organisation they work at, and those differ enormously.
In an organisation of thirty people, the system administrator is also the service desk, the buyer, the network administrator and the one who knows the phone system. In an organisation of three thousand they are one of twelve and do one thing well. Same title, different job.
On top of that, titles move with what is on the market. "Cloud engineer" did not exist fifteen years ago and now describes work that then sat partly under system administration. That makes old job adverts misleading: they ask for a title that has since come to mean something else.
What does hold up is the question of what someone does with their day. So below, each role is described by the work, not by what belongs on the business card.
The administration roles: what keeps running
These three keep upright what is already there. The work is cyclical, the result is that nothing happens, and that makes it undervalued until it goes wrong once.
- System administrator. Servers, virtual machines, operating systems, updates, accounts and rights, backups and the monitoring on them. They make sure the foundation stands. In smaller organisations the same person also does the workplaces and the service desk.
- Application manager. Responsible for one or a few applications: the ERP, the patient record, the planning package. They keep the application working, put changes through, test the supplier's new version and are the point of contact for the people using it.
- Network administrator. Switches, firewalls, Wi-Fi, the links between sites and the traffic across them. A role that disappears into system administration in small organisations and becomes a craft of its own from about five locations upwards.
Around application management runs one distinction that causes a lot of confusion: technical, application and functional management. Technical management is about the machine it runs on, application management about the application itself, and functional management about what the organisation wants from it — which fields, which rights, which process. That third one is often not an IT job at all but sits with the department using the application. Advertise for an application manager when you mean functional management and you get the wrong applicants.
The service desk: first line is not lowest line
The service desk is where the organisation meets IT. The work is often described in lines, which is useful as long as you know it refers to the kind of question and not to the level of the people.
- First line. Takes it in, resolves what can be resolved directly, and makes sure the rest is complete enough to pass on. This is where most of the influence on how IT is experienced sits, and that is no small responsibility.
- Second line. Picks up what needs more time or more rights: the workplace that is broken deeper down, the application behaving oddly.
- Third line. The specialists, and in practice often the vendor of the package itself.
The common error of thinking is that the first line is a stepping stone to the real work. Sometimes it is, but someone who is good at it — staying calm, asking the right question, knowing when to hand something on — is harder to replace than someone who can install a server. How we set up those lines sits under IT service desk.
The building roles: what does not exist yet
Here the weight is on putting things down rather than keeping them running. Different work, different rhythm, and usually a different kind of person.
- Cloud engineer. Builds and manages environments at a cloud provider: networks, machines, rights and the automation around them. The line with system administration is thinner than it looks; the difference is mostly whether the environment is put down by hand or in code.
- Data engineer. Makes sure data gets from the systems where it arises to a place where something can be done with it: reporting, analysis, a model. They build the pipes, not the conclusions — that last part is the work of an analyst or data scientist, and those two get confused regularly.
- Security specialist. A collective name that breaks into at least three jobs: someone who monitors and responds to incidents, someone who tests whether it can be broken, and someone who sets up policy and evidence. They do not ask for the same person.
These roles are often hired with a maintenance brief attached, because the organisation cannot yet keep the work apart. That goes well for a while, and then the builder leaves.
Which role you actually need
Before you put a title on a vacancy, three questions usually give you the answer already.
- Which work is not getting done? Not "we need someone", but: which tasks are currently not happening. Write them down. Often it turns out to be two things belonging to different roles.
- How many hours a week is that? Under twenty hours a permanent hire is rarely the best shape, and you are closer to contracting or outsourcing. Above thirty it tips.
- Is it cyclical or one-off? A migration is a project and has an end; administration is a state and does not. For the first you contract in, for the second you hire or outsource.
The trade-off between recruiting and contracting sits in more detail in IT staffing or hiring. If the first question shows it is mostly cyclical maintenance, outsourcing is a third option; what that involves sits under managed IT.
Questions we get about this
The ones that come up most often when this is on the table.
What exactly does a system administrator do?
They keep upright what sits underneath your workplaces: servers and virtual machines, the operating system on them, updates, accounts and rights, backups and the monitoring that says something is breaking. In a small organisation the same person also does the service desk, the workplaces and the purchasing; in a large one it is a bounded role. That is why two vacancies with the same title can be asking for very different people.
What is the difference between technical, application and functional management?
Technical management is about the machine and the platform an application runs on. Application management is about the application itself: keeping it working, putting changes through, testing new versions. Functional management is about what the organisation wants from the application — which fields, which rights, which process — and that role often does not sit in IT but with the department using it. Advertise for an application manager when you mean functional management and you get applicants who are good at the wrong thing.
What does a data engineer do?
They make sure data gets from where it arises to a place where something can be done with it: a report, an analysis, a model. That is building work — making connections, converting formats, checking it is right and making sure it happens again every night. The difference from a data analyst is that the engineer builds the pipes and the analyst draws the conclusions. Those two regularly end up in one job advert, and then nobody feels at home in both halves.
Do we need our own administrator, or can it be outsourced?
That depends on two things: how many hours the work is and how specific it is to your organisation. Generic cyclical maintenance — updates, monitoring, backups, accounts — outsources well and benefits from a team that is also there at night. Work entangled with your applications, suppliers and people belongs inside. The combination we most often see work is exactly that: one of your own people for the specific, and the generic with an on-call rota moved outside.
Where this lands with us
The services this subject falls under.
Not sure yet which role you need?
Name the work that is not getting done and roughly how many hours that is. Half an hour is usually enough to tell whether you recruit, contract in or outsource.
Practical IT knowledge in your inbox
New guides on management, security and the workplace, written by the people doing the work. No sales talk, and you can unsubscribe in one click.