Two Hangars, Two Security Habits: ThreatOps Compared

Safety people in aviation talk about three ways of learning: reacting to events that already happened, finding hazards before they bite, and predicting where trouble will appear next. Security teams face the same choice, and ThreatOps sits firmly in the second and third camps. To make the contrast tangible, here are two imaginary training operators, which I will call Hangar One and Hangar Two. Both have a modest IT footprint, both worry about attackers, and they handle the worry very differently.

Hangar One: the alert-and-block approach

Hangar One buys three threat feeds because a conference talk said more sources meant more safety. Every list of suspicious addresses flows into the firewall and mail filter with no review. Early on, the dashboards look impressive: thousands of indicators ingested, dozens of weekly reports circulated, a colourful coverage chart on the office wall.

The trouble shows up in small ways. Staff complain that a booking page cannot be reached from a partner airfield. A scheduling tool stops syncing. When someone finally investigates, they find that an automated block had caught an address inside shared infrastructure used by thousands of harmless sites. Nobody is sure which feed caused it, because nobody graded the sources or set an expiry date, so entries linger long after they stopped meaning anything.

Then a real event arrives. A login that looks perfectly valid is used from far away, and a second phone is quietly registered against the account. No indicator in any feed matches, so no alarm fires. The team learns about it from a customer.

Hangar Two: the question-and-hunt approach

Hangar Two starts smaller. It names one person as owner and writes five questions, each with a single decision attached, for example whether remote access needs a second check before the autumn intake. It lists the handful of sources it follows and scores each twice, once for the provider’s history and once for the claim in front of it. Entries get expiry dates by type, shorter for network addresses and longer for file fingerprints.

Feeds there start in alert-only mode. The operator reads the same guidance the Ops Atlas quotes from FIRST, that feeds are for flagging rather than unattended blocking, and adds a gate of its own before anything is blocked: a top-grade source, a credible claim, a hit rate of nine in ten over a quarter, and no match against the allowlist. Anyone on call may reverse a block, and the owner reviews it the next hour.

Every second week the team runs a small hunt on one behaviour. One of them tests the same story Hangar One suffered: valid login, odd country, new device registered. The logs go back long enough, the search finds nothing, and the analyst writes a note that the behaviour would have shown if it had happened. A tidy side effect is a request to the alarm owners to turn the search into a permanent rule.

Side by side

  • Starting point. One begins with sources; the other begins with questions.
  • Trust. One trusts every list equally; the other grades them and expires them.
  • Blocking. One blocks by default and apologises later; the other earns the right to block with measured precision.
  • Looking. One waits for alarms; the other searches deliberately and records negatives.
  • Measuring. One counts volume; the other counts decisions changed.

The Atlas draws a similar line. Its entry, which you can read as the Ops Atlas description of ThreatOps, lists “more feeds, more protection” and hunting as overflow for the alert desk among its anti-patterns, and warns against treating a heat map of covered techniques as the goal.

What each hangar reports at quarter end

Hangar One reports totals: indicators ingested, reports published, hunts started. The Atlas dismisses all three as vanity measures. Hangar Two reports ratios, modelled on the worked quarter in the Atlas: how many questions got answered, how much intelligence prompted action, how many hunts bore fruit, how many gaps were closed. In the Atlas’s example those come to 83, 35, 42 and 44 percent, and one feed that raised 120 alerts of which only 18 were real stays alert-only with its renewal under review.

Notice that Hangar Two’s numbers are not flattering. That is the point. A figure of 35 percent says where to improve; a pile of indicators says nothing.

Why the quiet result matters

The sharpest difference is how each treats an empty finding. Hangar One reads silence as safety. Hangar Two treats silence as a claim that must be supported by good data. The Atlas points to a warning that American agencies issued early in 2024 about a state-backed group whose presence in some networks lasted years, with conventional indicators described as thin on the ground; sweeping for known items and then declaring “not affected” would have been the wrong conclusion. Behaviour, not lists, was the better guide.

It also cites the industry-report median of 14 days of dwell time in 2026, up from 11. Two weeks is a long time for an unnoticed guest, and it is the window a hunting rhythm of one session a fortnight is designed to cover.

Does Hangar Two need a big budget?

No. The Atlas adds that a lean team can get going with shared notes kept under version control, no platform purchase needed, and its decision tree is practical: if you lack ninety days of usable logs, fix that first; if nobody will own findings, buy hunting as a service; if no one will ask questions, keep an advisory model; otherwise build. The first thirty days cost an owner, five questions, a source list, expiry rules and a place in an information-sharing group.

Which one are you?

Most operators are a blend, which is perfectly normal. The Atlas offers a ladder that begins with nothing in place, climbs through feed consumption in alert-only mode and a question-led register, and reaches regular hunting and finally intelligence-led operation. A useful exercise is to place your own organisation honestly on it, then choose the single next change. For many, that change is simply turning one automatic block back into an alert, and writing down the first question you want answered.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top