Neither the team nor the hospital below exists; both were made up for the telling, and no real organisation is meant. The pattern, though, is familiar: an administrative office with forty-odd laptops, a growing pile of tablets and an unwritten rule that “somebody in IT handles it.” We follow that fictional office through one quarter of EndpointOps, the informal practice of managing and securing staff devices as a single life-long routine. The story stays on non-clinical administration; no care, treatment or patient information appears anywhere in it.
The structure of the quarter is adapted from the plan outlined in the Atlas page on EndpointOps, which proposes a thirty-sixty-ninety day progression. Where we add detail, it is invented for illustration and labelled as such.
Day one: an honest count
Our imagined coordinator, who runs the office, opened a Monday meeting with a plain question at a Monday meeting: how many devices do we have? Three people gave three different answers. Purchasing records said forty-four laptops. The management console listed thirty-six. The protection software reported thirty-one. Nobody had been careless; each source simply had a different history. Analytically, the discrepancy is the finding. The Atlas treats this three-way mismatch as the signature of the early maturity levels, where a team has enrolled devices but has not yet reconciled the sources.
The coordinator did what record-keepers always do. She created one sheet with a row per device and marked every row present in some sources and absent from others. Eleven rows turned out to need attention: five machines nobody could find, four never registered, and two whose protection agent had stopped reporting. Record-keeping habits like these are the subject of a down-to-earth piece on our network, a practical farm budget book and money habits article, which argues that the simple act of writing every figure down beats clever tools.
Days two to thirty: standards and defaults
In the first month the team wrote a device standard of six lines and checked the platform’s defaults. The fictional administrator found what the Atlas warns about: an out-of-the-box setting that treats devices without an assigned policy as compliant. Four of the unregistered tablets had been passing access checks purely because nobody had given them rules. She flipped that option so machines lacking any rule would fail instead of pass, and the dashboard turned red for the first time. It was uncomfortable and useful.
She also moved her day-to-day work off the all-powerful administrator login. A narrower role handled routine work, and the powerful account was reserved for emergencies, with its credentials stored away.
Days thirty-one to sixty: keys and pilots
The second month focused on recovery keys. The team discovered that nine encrypted laptops had no stored key, a gap that would turn a forgotten password into a lost machine. They backfilled the keys until more than ninety-five percent of encrypted devices were covered, which is the target the Atlas gives for this phase. They also learned that deleting an old-looking record could remove a key, so nobody was allowed to delete a device without a second person’s agreement.
Enforcement began with a pilot of six laptops. For application control, the team ran audit mode first and reviewed what would have been blocked. Two scheduling plugins that staff genuinely needed appeared on the list, and approvals were added before anything was enforced. Skipping that audit is a recognised anti-pattern in the Atlas, and the pilot showed why.
Days sixty-one to ninety: wider enforcement and a drill
By the third month the office was ready to widen enforcement. A security group with a few trusted members was created as an approver group for destructive actions, so that any mass wipe would need a second person’s consent. The coordinator tested it with a harmless retirement of an old tablet, noting that nested groups would not have worked and direct membership was needed.
The quarter ended with a drill. Someone picked a random laptop and the team timed how long it took to retrieve its recovery key. The first attempt took twenty-two minutes, mostly spent finding who had access; the second, a week later, took four. It was this timing that the coordinator reported upward, not a statement of virtue.
A note on cost and effort
Readers who manage budgets will want to know what the quarter demanded. In the story, the effort was roughly two half-days a week from one person, plus short sessions from three colleagues. Direct spending was close to zero, because the tooling was already licensed. The one purchase was replacement hardware for two laptops running an operating system that no longer received security fixes. The Atlas puts a price on the alternative: paid extended support begins at $61 per device in its first year and doubles each year, so patience is expensive.
The invisible cost was attention. Several tasks only worked because someone owned them by name, and the coordinator learned to write the owner’s name on every row of the plan. Without that, tasks drift, exactly as devices do.
Analysis: what the story shows
Three observations follow from the invented example, and they match the general guidance. First, the biggest gain came from the first week’s reconciliation and cost almost nothing. Second, tenant defaults mattered more than any new tool. Third, the drill converted vague confidence into a number that could be improved.
The same coordinator also tracked budget exposure. Her team used a simple monthly spreadsheet along the lines suggested in a seasonal budgeting plan for gifts and bills, listing devices that would soon fall out of vendor support so that replacement costs were never a surprise.
A final detail deserves mention. At the end of the quarter the coordinator held a short, blame-free review in which each person said what surprised them. The recurring answer was that most problems were administrative gaps, such as an unowned task or an unrecorded handover, and not technical faults. That insight shaped the next quarter’s plan more than any dashboard did, and it cost nothing beyond an hour of conversation and a shared page of notes.
Limits of the story
An invented office proves nothing statistically, and real organisations differ in size, tooling and regulation. The value of the narrative is that it shows the sequence: count, standardise, fix defaults, backfill keys, pilot, enforce, drill. Teams in different sectors can swap the details while keeping the order. If you try it, record your own numbers at the start and at the end of the quarter, and let the comparison, not the story, make the case.