Administrative offices that support any busy facility keep long lists: rotas, supplier contacts, badge holders, shared mailboxes. The digital cousin of those lists is the set of accounts people use to reach scheduling, messaging and document systems. Tending that set is called IdentityOps, and this step-by-step guide shows an office team how to begin. It is a general operations explainer drawn from the Ops Atlas, intended for back-office staff; it does not discuss patients, treatment or clinical systems, and it is not professional advice of any kind.
Step 1: decide which record is the master list
Before touching any software, choose the document that says who works here. For most offices it is the personnel or rota record kept by the staffing team. The Atlas calls changes to this list joiner, mover and leaver events, and recommends that account changes flow from it instead of from ad hoc emails. Write the name of the list, its owner and the way changes reach the technical team on a single page.
Step 2: list every system and who answers for it
Build a plain grid, one line for each application: appointment scheduling, staff messaging, shared drive, payroll portal, and so on. Add the person responsible and where the user list can be exported. This register is the foundation of everything that follows, because you cannot clean what you have not found. Do not forget small tools that someone signed up for years ago on a work email.
Step 3: compare the lists and note the strays
Export user lists and set them beside the master record. Mark anyone who appears in an application but not on the roster. Mark logins named for a function rather than a person, because a shared login is a sign that nobody can be held to account for what it does. The Atlas names shared service logins among the habits to retire. Treat the strays as questions, not accusations; many have an innocent history.
Step 4: design the departure sequence with a stopwatch
When somebody leaves, a good routine changes the master record, then disables the central account, then ends open sessions, then tells each connected application to remove the person. Add a timestamp to each stage. The Atlas gives sample targets of a few hours for the central account and a full day for everything else. Practise on a pretend leaver and keep the notes. There are quirks worth knowing: with the standard provisioning protocol, an “inactive” flag has a meaning each application decides for itself, so verify what yours does rather than assuming.
Step 5: tighten recovery before it is tested
Write a short procedure for staff who lose access. Offer a self-service path using a second method registered in advance. For assisted cases, have a supervisor confirm the request before a temporary pass goes out, and never accept a request that arrives only by an urgent-sounding message. Keep a log of every reset. The Atlas stresses that the weakest door in many organisations is the one marked “I forgot my login”, which is exactly why a plan matters more than a tool.
Step 6: set a rhythm and measure honestly
Put four recurring tasks in the calendar. Daily: compare the master record with the account list. Weekly: sample some recoveries. Monthly: review automated accounts nearing expiry. Quarterly: run a departure drill and review outside sign-in services you rely on. Report completion across all systems rather than the central one alone; in the Atlas’s example, a figure of 90 percent shrinks to 82.5 once lingering application accounts are counted.
A word on timing for small teams. If a daily comparison feels heavy, start weekly and shorten the gap as the lists become cleaner. What matters is that the comparison is on the calendar, has a named person, and leaves a note behind, even if the note says that nothing changed.
What the first ninety days should leave behind
By the end of the third month a modest office should hold a handful of tangible things: a named owner for every tool; a register of applications; stronger sign-in, such as hardware keys, for the few who can change payment details or system settings; one staff group whose accounts follow the master list automatically; an inventory of automated accounts with owners; a recorded departure drill; and a one-page recovery procedure. The Atlas lays out this sequence in thirty, sixty and ninety day blocks. None of it needs a large budget, and each item produces evidence you can show a colleague or an auditor.
Common mistakes in an IdentityOps rollout
- Buying before cleaning. The entry cautions against purchasing a governance suite before staff data is tidy.
- Counting weak methods as strong. A text code or tap prompt is not in the same class as a hardware key.
- Forgetting the software accounts. Automated connections need owners and expiry dates too.
- Relying on first-login account creation alone. It offers no way to remove someone later, so combine it with a proper provisioning feed.
Where to go from here
If you enjoy building structured routines, you may also like this piece on a practice plan for DevOps interview rounds, which approaches skills the same stepwise way. And for a household-scale parallel about keeping tidy records, see the post on planning a first salary with a money-tracking app.
The Ops Atlas keeps a long reference that gathers definitions, standards and example cases; it is available as its IdentityOps chapter. Because definitions in the field are still unsettled, adapt the steps above to your own policies and check any change with the people responsible for your organisation’s compliance.