This is a fictional story, assembled to illustrate an operations idea, about a fourteen-person team that prepares document packs for people applying for residence and study permits. The team, the numbers inside their story and the people in it are invented. The ideas they stumble through come from a real reference, the Ops Atlas, and from public guidance it cites. Nothing here is immigration advice; it is a narrative about keeping the accounts of a document-handling business in order, a habit the Atlas groups under the name IdentityOps.

The Monday that started it

Our fictional team ran a client portal where applicants uploaded passport scans, offer letters and bank statements. On a Monday in spring, a coordinator doing routine housekeeping noticed an unfamiliar name in the portal’s user list. It belonged to a freelance translator who had finished her contract five months earlier. The account was active, and it had read access to every open case folder.

Nothing was misused. But the coordinator, whom the story will call the operations lead, spent the next hour working out how the account had survived. The answer was dull: when the translator left, the office manager removed her from the company email system and from the team chat. The portal kept its own user table, and nobody had told it.

What the evidence says about that kind of gap

The lead went looking for a framework and found the Atlas’s long entry on the subject. You can read the same material in the Ops Atlas discussion of IdentityOps. Its central claim for a story like ours is that a disabled login and a completed departure are different things. An organisation can switch off the primary account and still leave copies behind in applications that hold separate user lists, and it can leave live sessions running even after the account is off.

The Atlas also gives a measurement lesson that the team would later copy. Take forty people who left. If thirty-six had the central account disabled inside the target time, the headline reads 90 percent. If three of those thirty-six still had an account in some lesser application, the true completion is thirty-three out of forty, which is 82.5 percent. The team decided to track the second number.

Chapter one: counting what they had

The first job was a census. The team listed every application that touched client documents. They expected six and found eleven: the portal, a scheduling tool, a form builder, an e-signature service, two shared drives, a translation-job board, an invoicing package and three others. For each one, they wrote down who administered it and whether it kept its own users.

That list is the equivalent of the first step in the Atlas’s practical path: owners named and an application register built within the first month. The register took eleven working hours spread over a week. It also produced the first real finding: four applications had no named administrator at all.

Chapter two: the export that told the truth

Next, the lead exported the user list from each application and compared it with the current staff and contractor roster. The Atlas is specific on method here: take the export from the application itself, not from the log of what a provisioning tool claims to have done, because the log records intentions and the export records outcomes.

The comparison produced nineteen accounts that matched no current person. Eleven were former staff or contractors. Five were test accounts. Three were nobody could identify. The team disabled all nineteen, waited two weeks to see whether anything complained, and deleted them with a note recording the date. It was the same disable, wait, delete order that the Atlas recommends for retiring credentials.

If you are curious how a team documents requirements before acting, our overview of the employment residence route listed for Sri Lanka shows the sort of structured checklist the story’s team aspired to apply to their own accounts.

Chapter three: the help-desk problem

A few weeks later, an applicant-support agent received a call from someone claiming to be a senior colleague locked out of her portal login. The caller was friendly, knew the names of two current cases and asked for a second-factor reset. The agent, anxious to help, started the reset and then paused because something felt off. She escalated. It turned out the colleague was in a meeting down the hall and had not called.

The story’s team now rewrote its recovery practice using the Atlas’s graduated approach. Ordinary lockouts use a second method already registered. Harder cases hold back the temporary code until the employee’s line manager has vouched for them face to face or over live video. Administrator accounts need supervised proof plus a second administrator. Calling back a number on file was dropped altogether, since the Atlas calls that route unreliable once phone numbers can be hijacked. Each recovery now sends a notification to the account owner.

Chapter four: the strong-sign-in gap

The team believed it had two-step sign-in everywhere. Its dashboard said 93 percent of staff were enrolled. But the lead noticed that most enrolments used text-message codes. The Atlas explains why that matters: in its strict reading the phishing-resistant category contains physical keys and PKI cards only, whereas texted codes and one-tap approvals have been defeated in real attacks.

They bought security keys for the three portal administrators first, then for the six people who handled payments and email, and finally for everyone. That rollout order mirrors the Atlas suggestion of starting where compromise would hurt most. Three months on they tracked two figures: the share of people with a strong method registered, and the share of real sign-ins that used one. The gap between those two numbers told them where policy was still letting weaker routes through.

Chapter five: the quiet robots

The last discovery was the least dramatic. A form on the website sent each submission to a spreadsheet, and from there to the portal, using a token created two years earlier by a developer who had since moved on. The token had broad access and no recorded owner.

The team adopted the Atlas’s lifecycle for machine accounts: record an owner and a purpose, add an expiry and a review rhythm, favour credentials with brief lives, rotate to a schedule, and retire with proof. They replaced the old token with a dedicated automation account whose credential expired every ninety days and whose owner was the operations lead. A monthly report would now warn them before it lapsed. For a deeper look at how a student route lays out its requirements stage by stage, see the page on the student residence category for Sri Lanka.

Where they landed after ninety days

By the end of the first quarter, the fictional team had done what the Atlas’s thirty, sixty and ninety day outline describes in broad strokes. Owners and the register came first. Then a feed from their staffing records for one group of people, plus automated removal in the five applications that supported it, and the machine-account inventory. Last came session cancellation added to the departure steps, a first full drill and alerts for password resets.

The drill gave them numbers. A pretend leaver was removed from the central account in forty minutes and from all eleven applications in two working days. Their own target was one day. Two applications needed manual removal and moved to the top of next quarter’s list.

What the story is meant to show

The narrative is invented, but the lessons are sturdy because they match the Atlas’s reading of public incidents: failures tend to defeat a process, not the cryptography. A forgotten user table, a friendly voice on a phone, a weak second step and an unowned token are all the kind of weakness that a checklist and a calendar can remove. The Atlas itself concedes a tiny outfit might skip a dedicated practice, and it adds that purchasing a governance product before the underlying records are tidy only speeds the confusion. The fourteen-person team needed neither. It needed a register, a few owners and the habit of testing a departure every quarter.

0 0 votes
Article Rating
Subscribe
Notify of
guest
0 Comments
Oldest
Newest Most Voted