A practice runs on records: filings, deeds, advice notes, correspondence, all of them expected to exist, be findable and be intact when needed. The systems holding those records deserve the same discipline the practice applies to its own files, and the name the technology world gives that discipline is SecOps. The term sounds like something only a security vendor’s counsel could love, but stripped of jargon it describes a simple arrangement between the people who secure your systems and the people who run those systems day to day. It is a bargain you are entitled to ask about, in plain language, at your next review meeting.
SecOps in one paragraph, no jargon
Here is the definition worth keeping. SecOps, in its most useful sense, is a working contract joining the security side to the operations side: the hardening standard every machine must meet is written down, every weakness found lands in one shared list, every item on that list has a team responsible for it and a date attached, and the numbers telling you whether any of this is happening are produced from the same list rather than from two competing spreadsheets. That is the reading the plain-language SecOps guide at the Ops Atlas adopts, and it deliberately excludes two other things the word is sometimes used for — the monitoring centre where analysts watch alerts, and a Google product of the same name. Neither of those is a contract; both are things you can buy or staff, which is exactly why they cannot substitute for one.
The three duties your IT provider owes
Translate the contract into what a non-specialist client should expect, and three duties emerge. They are few, and each is checkable.
- Keep the standard written and current. There should exist a document saying which settings your servers, laptops and cloud services must have — and it should carry a version and a date, the way any precedent does. “We keep things patched and safe” is a reassurance; a named standard is an obligation.
- Fix findings by owner and date. When a weakness is discovered, it goes on one list with a responsible team and a deadline the team can meet. A list without dates is a wish, and a list scattered across several trackers is several wishes.
- Put exceptions in writing, with an end date. Sometimes a fix must wait — a supplier system nobody may touch, a case-management platform locked by the vendor. Legitimate delays happen. What distinguishes a managed exception from a quiet abandonment is the same thing that distinguishes an adjournment from a dismissal: paperwork, an owner, and a date it comes back to court.
The records your systems must keep
Lawyers understand custody, so the telemetry half of the arrangement translates naturally. Industry guidance on cyber defence sets out what records the operational side of a business owes to whoever watches its security, and the duties scale with the size of the practice. At the basic tier: a defined process for logging, audit logs actually collected on your systems, and somewhere to store them. At the next tier: clocks synchronised across machines so timestamps agree, logs gathered centrally rather than scattered per device, retention of at least ninety days, and a weekly human review of what was collected. Two practical consequences follow. If an incident is investigated, the evidence chain is only as old as your retention; and if your provider cannot state any of the above in a sentence, the duty has no owner.
Names worth knowing when reading proposals
Proposals from IT firms borrow vocabulary freely, and a little recognition goes a long way. The CIS Controls, currently at version 8.1, are a prioritised to-do list of defensive measures, widely cited in questionnaires. The NIST Cybersecurity Framework, in its 2.0 edition, organises the same world into functions governance teams recognise. The CISA KEV catalogue lists flaws known to be actively exploited — when one appears on your systems, the clock is short and the excuses few. Beneath these sit the detailed publications: configuration checklists, configuration management and patching guidance from NIST’s Special Publication 800 series, and the government-grade hardening standards known as STIGs. A provider who can map their service onto those names is describing an arrangement; one who cannot is describing sentiment.
Questions to ask about your SecOps arrangement
Bring six questions to the next review. Honest answers take minutes; evasion takes longer, which is itself informative.
- Which hardening standard do our systems follow, and what version of it?
- Show me the list of open weaknesses. Who owns each item, and when is each due?
- How many items on that list have no owner at all this week?
- Which exceptions are in force, when does each expire, and who reviews them?
- What share of our machines send their logs in, and where are they kept and for how long?
- What did a security change break in the last year, and what changed in your method afterwards?
One warning about tooling lock-in
A caution worth raising before signing long monitoring contracts: tools move houses. Microsoft has announced that Sentinel, its cloud monitoring product, departs the Azure portal once 31 March 2027 has passed — a reminder that even flagship security platforms get relocated by their makers. The lesson is not to avoid tools but to ask the portability question while negotiation is easy: if this platform changes or we change providers, do our logs, rules and records leave with us in a usable form? A practice that keeps its case files portable would accept nothing less from its security records, and the same instinct serves here.
None of the above turns a lawyer into an engineer, and none of it needs to. The arrangement described here is, in the end, a matter of competent administration: a written standard, one honest list, dated exceptions, records kept to a stated term. Practices already trust each other with such things daily. Your systems are owed the same courtesy, and asking six questions is a fine way to collect it.
