A freelancer who takes over part of a client’s technology operations inherits more than a task list. You inherit their habits, their unwritten rules and their blind spots, and when something breaks at midnight, the contract will say you were responsible. Before signing, a short ITOps audit tells you what you are really buying. This checklist is built for solo consultants and small contractor pairs, and it asks for evidence at each step rather than assurances.
The vocabulary here is borrowed from xOps School’s Ops Atlas. Reading the ITOps entry, you find the discipline defined by five duties: handling requests, handling incidents, fixing root causes, controlling change and maintaining an estate map. If a client cannot show you any of those five, you are being asked to build an operations function for the price of a support retainer.
How to use this audit
Give every item a score of 0 (nothing exists), 1 (exists but informal) or 2 (written and used). Collect the evidence in the first call or two, and record the source for each score, whether a screenshot, a document or a shared link. A total below 6 of 10 means the engagement needs either a paid setup phase or a narrower scope. A total of 8 or more means you can mostly just do the work.
The approach follows a simple evidence-first principle: claims count only when something concrete backs them up. That is also how the Atlas treats its own sources, labelling the status of terms and noting when a case is a preprint that has not been peer reviewed.
Item 1: intake and ranking of work
Ask how problems reach the team today. Email chains, chat messages and phone calls are all common, and none is fatal, but a contractor needs one place where every item has a name, a time and a state. Check whether urgent items and routine ones are separated.
Evidence to request: the last twenty items, with their timestamps. Score 2 only if you can tell from the list which were outages and which were ordinary asks.
Item 2: who is allowed to change what
Request the written rule for alterations to live systems. In the Atlas framing, there should be a pre-agreed route for routine low-risk changes, an assessment path for the rest, and an emergency route that is reviewed afterwards. As a freelancer, you want to know which route you may use alone and which needs someone else’s sign-off.
An important point for people working under a contract: the Atlas says regulated financial firms in the EU face expectations of an independent approval function. If your client falls in that bucket, you may not be the person who approves your own emergency fix.
Item 3: the risk score
Ask how the client decides which changes need a second opinion. Anything written is better than nothing. In an Atlas teaching example, a change reached a score of 6 from three components: the tier of the system, an alteration to a database schema and a start time outside the agreed window. That total triggered a board review. Ask whether your client has any equivalent, even a three-line version.
Item 4: response clocks and what they exclude
If the client has promised response times to its own customers, find out the exact start, stop and pause rules before you accept any share of them. The Atlas shows the stakes with a lab ticket that spent 92.5 hours open yet consumed only 4.5 hours of its allowance, thanks to weekend and holiday pauses. The same exercise shows another danger: a promise to a customer is built from smaller promises inside and outside the company, so your own response time becomes one block in someone else’s sum.
Write down your own allowance in the contract. If a customer commitment is eight hours and you are asked for a diagnosis window of two, check the sum of all pieces, including the buffer the client has quietly left for itself.
Item 5: the estate map
Ask for a list of systems, accounts, renewals and owners. Count the rows that are stale, orphaned or duplicated. In the Atlas health example, ten records produced two stale entries, one orphan and one duplicate, a ratio you can reasonably expect from an unmanaged list. A fair list for a small client might be one page long.
Mention to the client that a register works when it holds what decisions require, rather than every device ever bought. A short, current register is a better sign than an enormous, outdated one. Also check for shadow spreadsheets: if three different people keep their own copy of the “real” list, score the item 1 at most.
Item 6: alerts and on-call
Freelancers are often handed an on-call phone. Ask who gets the first alert, what happens if that person does not respond, and how the alerting itself is tested. The Atlas lists the event layer treated as non-production as an anti-pattern, because a silent failure can hide an outage.
A practical wrinkle: the Atlas notes a retirement date, 5 April 2027, for the incident-paging product from one large software company. If the client relies on it, ask for the migration plan, and make sure moving the alert routing is not silently added to your scope.
Item 7: reporting duties
If the client falls subject to the European security directive abbreviated NIS2, the Atlas summarises three deadlines: an initial alert after a day, a fuller notice at three days, and a closing report a month on. A contractor who sees a serious incident needs to know the internal route for escalation, because the clock is the client’s but the discovery may be yours. This is general information, not legal guidance; ask the client’s own counsel how it applies.
Item 8: review habits
Ask whether failed changes and emergency changes are reviewed after the fact. The Atlas lists the review of every such case as a core mechanic. If a client says reviews happen informally, request one example. A single written note about what changed afterwards is enough for a score of 2.
Item 9: tooling versus practice
It is tempting to judge maturity by the logo on the ticketing tool. The Atlas warns against that: the software records the practice, it does not constitute it. A client with a free or lightweight tool and clear habits often outperforms one with an expensive platform and no rules. Score the habits, not the product.
Item 10: your own capacity
The last item is about you. If you hold a day job as well, the guidance in our piece on freelancing responsibly while employed applies doubly to on-call work. Score yourself on whether you can answer within the client’s expected window on three random days next month.
Common objections you will hear
Clients sometimes push back on being audited by their own contractor. Three responses tend to work. First, the audit protects them as much as you, because it surfaces gaps before an outage does. Second, it is short: most items need a single document or a five-minute screen share. Third, it is repeatable, so a second pass six months later shows progress with numbers rather than impressions.
Another objection is that a small firm does not need any of this. Often that is true for the heavy parts. The point of the scoring is to show which light parts matter, and for many small clients the answer is two or three items rather than ten.
Turning the score into a scope
With ten items scored, write a one-page scope note. For every item that scored 0, either add a priced setup task or exclude it explicitly. For every item at 1, list the written artefact you will produce in the first month. It is also worth sending newer freelancers to the starter FAQ for people new to this line of work before they negotiate their first operations retainer.
One caution runs the other way: do not impose a heavyweight framework on a very small client. The Atlas names heavyweight service management on a single-service team as an anti-pattern. For a five-person firm, the audit above plus a shared calendar and a one-page register may be all the structure that is warranted.
A final habit worth building is to repeat the audit every six months, because clients drift. Evidence-led scoring gives you a neutral record when a scope conversation turns tense, and it keeps the relationship focused on what is written down instead of what everyone remembers.