
Over the years I have heard the same half-truths about incident work repeated in meetups, interviews and chat groups, usually by smart people who have not yet had to run a bad night themselves. Anyone growing a DevOps career who hopes to become the name colleagues reach for on a bad night should know which of these beliefs to let go. Here are five myths about IncidentOps, with the friendlier reality behind each and a way to turn it into something you can show on your profile.
Myth 1: “Incident work is just being good at debugging”
Debugging matters, but what a catalogue entry from the Ops Atlas about IncidentOps sets out leans mostly on people and process: declaring, commanding, communicating and learning. It readily admits the name itself is casual slang for incident management, yet its emphasis is unmistakable. The best technical person in the room might be the worst coordinator, because a coordinator’s job is to ask questions and keep the picture, not to dive into a terminal.
Career move. Volunteer for a role that is not “the fixer”. Taking the scribe or communications seat on a couple of incidents shows range and teaches you how the whole response fits together.
Myth 2: “More action items mean a better review”
I have sat through reviews that ended with twenty tasks and left everyone feeling virtuous. Three months later most were untouched. The Atlas calls out the pattern: actions that are vague, equally weighted, partly just mitigation, or closed by quietly downgrading them. Its alternative is a short list where every item has one owner and one due date, plus a service-level target for finishing the important ones. One big software company, for instance, gives itself a month or two depending on the service.
Career move. After your next review, report the completion rate of its actions a month later. Few people do, and managers notice.
Myth 3: “Everyone should join the call, just in case”
It feels supportive, and it backfires. The Atlas cites field research describing large incidents where hundreds of participants pushed the core team into private conversations, and a documented cluster outage that had 41 people in the chat and seven separate task forces. A crowded room makes decisions slower and mistakes easier.
A leaner arrangement works better: a small circle of hands-on people, with updates broadcast in a second, listen-only room. A handy rule from emergency management, quoted by the Atlas, is one supervisor per five people, with sub-teams forming beyond that.
Career move. Be the person who suggests splitting the channel early. It is a small act that shows judgement.
Myth 4: “Being on call is simply part of the job, whatever the load”
Many careers have been dented by this belief. The Atlas runs a quick sum: a roster of six engineers covering every hour, two shifts each per week, works out near 33 percent of their time on duty, past the 25 percent cap that Google’s guidance sets. Reaching that limit sustainably calls for around eight people. In other words, burnout is not a personal weakness; it can be arithmetic.
The same entry mentions a 2026 community survey, with 1,363 self-selected respondents, in which 30 percent named alert fatigue as a pain point. Read it as colour, not proof; it still rings true to many of us.
Career move. If you are asked to join a rota, do the sums and raise them politely. Showing you understand sustainable cover is a mark of seniority, not complaining.
Myth 5: “A falling average repair time proves we are improving”
This one catches managers and engineers alike. According to the Atlas, the familiar repair-time acronym can refer to four or more distinct phases, and because durations are lopsided, a few long events set the average. A study by a Google engineer, as summarised there, found that even a real ten percent improvement could show up as a worse number in a substantial share of simulated quarters, while a team that changed nothing still had close to a 19 percent chance of appearing thirty minutes better.
The Atlas’s own lab example has eleven incidents with a mean mitigation time of about 78 minutes and a median of 32, and a single severe event accounts for most of the difference. Better to show distributions, the budget of acceptable errors used, pager load and the closure rate of actions.
Career move. Offer to rebuild one report from an average into a distribution chart. It is a fast way to look like the person who understands data.
What I would tell my younger self
Junior engineers often assume that dramatic rescues earn credit while quiet routines go unseen. Experience points the other way. The engineers I trust most are the ones who keep a checklist, say their role out loud, post updates on time and write short, honest reviews. They make the rest of the team better, which is what seniority really means. Should a single habit survive from this article, let it be the spoken handover: tell the next person plainly that command is theirs and wait for confirmation. It takes five seconds and removes an entire family of mistakes.
A bonus myth: “You need a special title to do this”
You do not. Roles in the described model are hats, not job titles, and whoever outranks the others during that particular call takes charge, whatever their normal standing. A mid-level engineer can lead a small incident and a senior one can take notes. If you want to grow, ask your team for a chance to wear a hat you have not worn before. Handing over is a ritual too: the outgoing coordinator states clearly that the newcomer is now in charge and waits for a yes.
A short script for the hard conversation
Careers also stall because people do not know what to say when a review turns awkward. Try these openers. When someone proposes naming a person: “What about the setup made that easy to do?” When someone suggests adding a dozen actions: “Which three would we regret skipping?” When someone asks why you did not wait for the cause: “We knew where the fault was, and undoing the change was the cheapest safe move.” Practising a few lines like these makes you steadier under scrutiny, and colleagues will remember how you handled the room as much as what you decided.
Putting it on your profile
Be concrete and modest. Rather than “incident management expert”, try a line such as “coordinated response to a customer-facing outage; introduced written declaration criteria; reported action completion rates.” Avoid inventing credentials. As the Atlas acknowledges, no formal certification ladder exists for this label. A simple blog post describing one lesson you learned, with details of the system removed, does more for your reputation than a long list of tools.
Where to practise
- Run a small tabletop exercise for your team with an invented failure.
- Write one-page role cards and test them.
- Rehearse your handover phrase with a friend.
- Keep your own mini-notes after any production hiccup, even a minor one.
If you work with containers, the habits fit nicely beside the roundup of typical beginner container mistakes and the guidance in practices that keep Docker steady from laptop to production, because many incidents begin with exactly such slips. Learn from the mistakes, rehearse your response, and keep the review friendly. That combination is rare, and it is what people remember when they decide who to trust next time.







