Owners of light aircraft compare options constantly: two engines, two avionics stacks, two hangar arrangements. The same habit helps with a less glamorous question. Your booking system, your maintenance records and your payment page all have weaknesses published against them, and you cannot fix every one at once. How do you decide the order? This piece compares three approaches side by side, using the framing in the VulnerabilityOps entry on the Ops Atlas, and then offers a verdict for small aviation operators.
A caveat first: the page presents the term as young, with no official standard defining it. What follows reflects that one source, and other sensible views exist.
Approach A: Severity first
The simplest policy says: fix the highest severity scores first, and treat anything marked critical as urgent. It is easy to explain and easy to audit, and that is why it is popular. The Atlas files it under anti-patterns once it turns into the entire policy, and its reasons are below.
Strengths: clarity, a built-in sort order, and wide tool support.
Weaknesses: the custodian of the severity standard says its base scores are not meant to be the sole basis for a risk call. A score describes how bad a flaw could be in general, not whether anyone is using it, and not how exposed your own system is. The Atlas also points to arithmetic: in terms of how many flaws need attention, treating critical-equivalent items as the target means chasing roughly the top tenth by probability, around a score of 0.04, while treating high-and-above as the target pulls in nearly half of all flaws, at a probability near 0.008. The effort rises sharply while the gain per fix falls.
Approach B: Patch everything
The second policy is the instinct of a conscientious owner: apply every fix, always. It sounds responsible and it feels fair.
Strengths: no judgement calls, no arguments, and strong intentions.
Weaknesses: capacity. A 2019 analysis covering around three hundred companies found the typical one fixing roughly ten percent of its open flaws each month, while the best performers managed 2.5 times as many. Meanwhile the flood of new records has more than doubled since 2020. For a three-person flying club, patch-everything usually degenerates into patch-whatever-the-vendor-emails, with no way to know what was missed. Its other drawback is that everything looks equally important, which means nothing is.
Approach C: Evidence plus clocks
The third policy, favoured by the Atlas, ranks signals in a fixed order: evidence of exploitation first, then probability, then severity. It derives a deadline from facts about the system and the flaw. The June 2026 US federal directive it discusses uses four: public exposure, a listing on the known-exploited catalogue, whether the attack can be automated and the technical impact. The result is a handful of tiers, each with a clock.
Strengths: urgency reflects real conditions; the top tier stays small; deadlines are explicit.
Weaknesses: it needs an inventory with exposure tags, a source of catalogue data, and someone to maintain the rules. It also relies on public data that has gaps. Since April 2026 the main US database fully enriches only prioritised entries, so some fields arrive late or never.
Side-by-side scorecard
- Effort to start: A is low, B is low, C is moderate.
- Effort to sustain: A is high because the top group is large, B is unsustainable, C is moderate because the top tier is small.
- Fit with real attacks: A is weak, B is accidental, C is strong.
- Explainability: A is excellent, B is excellent, C is good with a one-page rulebook.
- Audit trail: A is thin, B is thin, C is strong if closure needs proof.
A note on mixing signals badly
A tempting shortcut is to multiply a probability score by a severity score and rank by the product. The Atlas rejects this firmly, citing the standards body behind both, which names the move score laundering. Two unlike measures get blended into one tidy figure that appears exact yet explains little. There is a practical wrinkle too: the probability model moved to a new version in mid-June 2026, replacing the previous one, so any threshold you set deserves a fresh look whenever the model changes.
What about decision trees?
Between A and C sits a further option, structured decision trees. Two public versions exist, one from the US cyber agency and one from the CERT Coordination Center, each leading to a small set of outcomes. They suit teams that like a documented, repeatable path and do not want to maintain numeric thresholds. The Atlas treats them as a compatible alternative, not a rival.
How each approach behaves under stress
Policies look similar on a calm day and diverge when something loud happens. Suppose an advisory lands on a Friday afternoon about a widely used web component. Under A, the team checks the score, sees a nine, and starts working, even if none of its public systems run the component. Under B, the team adds it to an enormous queue and hopes. Under C, the team asks four questions, finds that one public site is affected and the flaw is on the exploited list, and starts a three-day clock before the weekend ends. The work is the same size in every case; only C aims it at the right place.
The difference shows up again at the monthly review. A and B can report activity, such as the number of fixes applied. C can report outcomes, such as the share of top-tier items fixed and verified inside their deadline. Outcomes are what an insurer, a partner or a customer cares about.
Verdict for small aviation operators
For a light-sport business with a handful of systems, approach C in its simplest form is the clear winner. Tag each system as public or internal, check the exploited-flaw catalogue each morning, use three tiers, and require proof of closure. You do not need heavy tooling, though if you are choosing software to support the routine, our guide to sorting through software options for your toolkit and the reminder that building tech skills step by step pays off will help you plan. If you already run a tight preflight regimen, the mindset transfers directly, as shown in our gear-by-gear inspection list for light sport aircraft owners.
Approach A remains useful as a tie-breaker within a tier, and approach B survives as the default for low-stakes internal items that get fixed at the next scheduled upgrade anyway. The weak spot to avoid is making either one the entire policy.
One metric to settle the argument
Arguments about policy tend to go in circles because each side quotes a different measure. A shared ratio ends that.
Whichever approach you pick, keep one ratio on the wall: new arrivals over verified closures, tier by tier. If the number stays under one, your policy is working for your capacity. If it climbs above one, you are accumulating a backlog that will eventually exceed every deadline you set, and the policy needs tightening or the capacity needs raising.

