Consultants who come in to speed up delivery tend to spend their first week staring at pipelines. A smarter first week looks at the operations function that has to live with whatever gets shipped. If the desk behind the platform cannot sort requests, track changes or explain its own response times, faster delivery just produces faster mess. This list gives eight ITOps checks that an engagement lead can run in the first fortnight, with a verdict each time on what “good enough” looks like.
For grounding, the Ops Atlas run by xOps School describes the discipline in its ITOps entry as the group tasked with keeping IT services alive day to day, covering requests, incidents, recurring-fault removal, change control and the estate map. That scope is the boundary for our checklist, and anything outside it can wait.
1. Can the client tell a request from an incident?
Ask to see last week’s tickets and sort them by hand. Routine asks with a known answer are requests; things that used to work and now do not are incidents. If everything sits in one undifferentiated pile, urgency is being guessed at.
Good enough: two lanes, named owners and a rule for ranking. The Atlas quotes one platform’s sample rule where priority equals impact plus urgency minus one. Any rule works if people use it consistently.
2. Is there a written change model, with limits?
Many clients proudly show a change policy that says “all changes need approval”. That is the first red flag, because treating every change alike is a named anti-pattern in the Atlas. What you want is a small catalogue: standard changes pre-authorised and low risk, normal changes assessed, emergency changes allowed to jump the queue.
Push further and ask whether any change recipe can lose its pre-approval. The Atlas gives an example of a model that gets suspended if it fails more than once in the last twenty uses. That kind of tripwire keeps a shortcut honest. If no recipe has ever been withdrawn, ask why not.
3. Does the risk score exist on paper?
Risk talk without numbers becomes opinion. A scoring sheet need not be sophisticated: a few factors, a weight each, and a threshold that sends a change to a review board. The Atlas worked example reaches a total of 6 by adding 3 for system tier, 2 for a schema edit and 1 for an out-of-window slot. Your client’s factors will differ, but the principle holds, and it lets a consultant argue about weights instead of personalities.
4. Is there one forward schedule that detects clashes?
Count the calendars. If the infrastructure team, the application team and the finance department each keep their own, collisions are a matter of time. The Atlas favours one forward-looking schedule that flags overlaps, and a lab where a payroll freeze between 26 and 31 October 2026 blocks overlapping work. One shared view is worth far more than any tooling around it.
5. What happens when the pipeline raises a change?
This one matters for anyone selling automation. The Atlas advice is that a delivery pipeline should create the change record but never approve it. The distinction is simple: the machine files the paperwork, a person or a policy rule decides. Check that approvals are not quietly self-granted by a build job, since that undermines the independence that auditors and regulators look for.
On the regulatory side, the Atlas notes that the EU’s financial-sector resilience rules require an approval function that is independent and approval after the fact for emergency changes. If the client is a bank or an insurer, confirm which rules apply before redesigning anything. Note also the research on software delivery says external approval boards show no link to lower failure rates and recommends peer review, which tells you the answer may be a blend. Keep the two DORAs straight in every workshop.
6. Does anyone review the failures and the emergencies?
Ask for the last three failed changes and the last emergency one. Was each reviewed afterwards, with a note about what to alter? The Atlas treats this review as a core mechanic, and for good reason: emergencies are where a recurring weakness likes to hide.
A related anti-pattern is piling on process after each failure. Resist the instinct to answer a bad Friday with a ten-step form. A review that changes one concrete thing beats a form that nobody reads.
7. Can they explain their response clocks?
Request every service target in plain language, including the trigger that starts the timer, the event that halts it and any days it ignores. The Atlas lab with INC-202 shows why this matters, because 92.5 hours of elapsed time counted as only 4.5 hours of target time after pauses. Clients who cannot explain their own clocks cannot defend them in a dispute.
Then check how the customer promise decomposes. An eight-hour promise in the Atlas example is the sum of a desk triage allowance, two diagnosis allowances, a carrier allowance and a buffer. If a client’s headline number is smaller than the sum of its parts, the number is a wish.
8. Is the inventory honest, and is it the right size?
Finally, ask where the estate map lives. A configuration database that tries to hold everything usually decays; one that holds what decisions need survives. The Atlas health example found two stale records out of ten, one orphan and one duplicate serial, which is a fair picture of typical rot. A plain spreadsheet used on the side is worth flagging too, as the Atlas lists shadow spreadsheets among the classic anti-patterns.
If you are planning a migration, this check matters twice over. You cannot move what you cannot list, a theme that runs through our complete guide to consulting on cloud migration. Treat the inventory as an exit criterion for the assessment phase.
What to ask the client to bring
The checks go faster when the client prepares a small evidence pack before the first workshop. Ask for a month of ticket exports, the current change policy, any calendar that shows planned work, the written definitions behind service targets, and whatever list currently stands in for an inventory. Do not be surprised if some of these do not exist; their absence is a finding in itself, and the honest answer “we do not have that” saves everyone an afternoon.
Explain at the start that the aim is not to grade the team. Operations staff are often wary of consultants because past reviews ended with a stack of new forms. Tell them that you are looking for the smallest change that removes a recurring annoyance, and that you will drop any recommendation they can show costs more than it saves.
Putting the eight checks into an engagement
None of this requires a certification programme. A sensible sequencing for a two-week discovery looks like this:
- Days 1 to 2: sample tickets and sort them into requests and incidents.
- Days 3 to 4: gather every change policy, calendar and approval path.
- Days 5 to 6: draft a risk score with the client and test it on ten past changes.
- Days 7 to 8: map response clocks and decompose each promise.
- Days 9 to 10: audit the estate register for stale, orphan and duplicate rows.
The output is a short findings note with three columns: what exists, what is missing, and what to fix first. Anyone reading our outline of an engagement from assessment through enablement will recognise where this slots in, before the enablement work begins.
A caution on scale
One more rule keeps consultants out of trouble. The Atlas warns against a heavyweight service management setup for a single team running a single service. If your client is a ten-person product group, the right deliverable might be a one-page change score and a shared calendar. Selling a framework certification to that audience is the easiest way to lose trust.
For larger estates, the Atlas lists options at different weights: ITIL 4 as optional guidance, ISO/IEC 20000-1 when certification is required, COBIT 2019 for governance and FitSM as a free, light option. Match the weight to the need, and write the reason down in the report.
Run the eight checks, score each as present, partial or absent, and you will have a defensible baseline before a single pipeline is redesigned. That baseline is what lets you prove, six months later, that the operations side improved alongside the delivery speed.
Best Cardiac Hospitals Near You
Discover top heart hospitals, cardiology centers & cardiac care services by city.
Advanced Heart Care • Trusted Hospitals • Expert Teams
View Best Hospitals