Book Appointment

The Quiet Hunt: A Hospital Group’s ThreatOps Story

Hospital IT

Everything below is invented: the team, the hospital group and the Tuesday in question. The working method behind it is real, and it is drawn from a catalogue entry that gets a proper citation shortly. The setting is a regional hospital group’s IT office, four people in a room with a whiteboard that nobody had cleaned since spring. Nothing in the story touches clinical work; it is about the machines that keep a group of hospitals running.

Their manager, a patient woman with a habit of asking “and then what?”, had read the Atlas write-up on ThreatOps over a weekend and arrived on Monday with one sentence on a sticky note: every hunt must end with something written down. The team shrugged. They had never hunted for anything; they answered tickets.

The question on the whiteboard

The note said to start with a hypothesis, not with a tool. The youngest analyst suggested one drawn from a news story about attackers who sign in with real stolen credentials and then add a second authentication method to keep access. “If that happened to us,” she wrote, “where would we see it?” The manager approved: one behaviour, one place to look, one fortnight.

The Atlas recommends a four-box frame called ABLE, standing for Actor, Behaviour, Location and Evidence, and the team treated it like a form to be completed. Actor: any intruder with stolen staff credentials. Behaviour: registering a new sign-in method. Location: the identity provider’s audit log. Evidence: a registration event with no matching helpdesk ticket.

Three days of looking

The quietest engineer pulled ninety days of identity logs and found the first surprise: the logs from the group’s two smaller sites only went back eleven days. Nobody had noticed because nobody had ever read them. He raised his hand. “This is where it gets awkward,” he said. Under the method, missing data is not an embarrassment; it is a finding. They opened a ticket titled “extend log retention at the two smaller sites” and carried on with the sites that did have data.

Across the remaining logs, every new registration matched a helpdesk ticket. The analyst checked twice, then a third time, and then the room went quiet. The team felt oddly deflated, as if they had searched a house for a leak and found only dry floorboards.

The argument about nothing

This was the moment that mattered. One colleague, a veteran of many audits, said they should shrug it off and call the hunt a waste. The manager asked a different question: “Could we have seen it if it had happened?” For the larger sites, yes: the data was complete, the query was sound, and a rogue registration would have stood out like a red coat in a snowfield. The Atlas has a term for this outcome, a confident negative, and treats it as a success rather than a miss.

For the smaller sites, no. There the honest answer was “we cannot tell,” and that answer was already captured in the retention ticket.

So the team wrote a one-page record in four short parts: the question, the data used, the result for each site, and the follow-up ticket. It also said, plainly, “we would have seen this at the three larger sites.” That sentence, the manager pointed out, was worth more to the board than a quiet dashboard.

The test of a stranger

The Atlas proposes a plain acceptance test: hand the write-up to a stranger and see whether they can reproduce the result. To check, the manager handed the page to a network administrator from a sister facility who had never seen the data. He ran the query in an afternoon and got the same result. Two lines of the write-up were unclear, so he fixed them. Only then did the team consider the hunt finished.

Afterwards, over stale biscuits, the team talked about why the exercise had felt so different from ordinary ticket work. Tickets arrive from outside and demand an answer by a deadline. The hunt came from inside, was shaped by their own curiosity, and finished when they decided it was done. Several of them admitted it was the first time in months that they had learned something about their own systems that nobody had asked them to find out.

What happened next

Three things followed, none of them dramatic. The log retention ticket was approved within a week, because it was no longer an abstract wish but the output of a real exercise. The team scheduled a second hunt for a fortnight later and picked a different behaviour. And the manager joined a regional sharing group for healthcare IT, where she learned the traffic-light labels used to mark how widely a note may be passed on. She now checks the label before forwarding anything.

The group also began tracking how many hunts ended in an artefact that changed something. That is a better measure, the Atlas argues, than counting how many hunts were started.

Takeaways from the tale

  • A hunt begins with a specific, answerable question and a limited scope.
  • Gaps in data are discoveries, not shortcomings.
  • A null result with good data is something you can say aloud with confidence.
  • If a stranger cannot repeat the work, it is not finished.

Further reading on this site

Teams who want the wider backdrop might enjoy the site’s primer on the foundations of dependable IT operations, and the broader essay that explains how several operational disciplines fit under one umbrella. Both help place a small hunting routine inside everything else an IT office must keep running.

The fictional team is still, as of this telling, a small one. They still answer tickets. But on the first Tuesday of every fortnight the whiteboard now has a question on it, and by Friday there is a page with an answer, or at least an honest account of why there is not one yet. This account is general and illustrative, and it is not a description of any real organisation.