Everything below is made up for illustration; no real company or customer stands behind it. It follows a small charter-booking team through one rough month and shows what ITOps looks like when it is applied sensibly. Names are left out on purpose; the roles are what matter.
The setup: a busy desk, a fragile system
Imagine a six-person charter desk. Two coordinators take enquiries, one handles pricing, one manages crew and aircraft availability, one does accounts and one person, the operations lead, also looks after the computers because she is the most patient with them. Quotes go out through a web form and a shared inbox. Availability sits in a scheduling tool. Payments run through a card service. None of it is exotic, and all of it is essential.
In March the form stopped sending notifications for about nine hours. Two enquiries sat unseen overnight. One customer went elsewhere. The operations lead restarted the mail connector, the notifications resumed, and everyone breathed out. Then it happened again twelve days later.
Turning point: the second outage
After the repeat, the lead did something the team had never done: she wrote it all down. Time of failure, who noticed, how, what she did, how long it took. She also borrowed vocabulary from the Ops Atlas entry on ITOps, which separates restoring service from removing the cause of the breakage. That distinction gave her a way to talk to the team. The restart was an incident fix; the real task was a problem investigation.
The Atlas calls the work of tracking down repeat-failure causes a central part of the function, and she liked that framing because it justified time she would otherwise have felt guilty about taking.
Week two: finding the cause
Reading through the two logs side by side, a pattern emerged. Both failures started shortly after a Tuesday-night plugin update that the pricing coordinator ran when it was quiet. The updates were harmless individually, but they briefly altered a setting that the mail connector depended on. Nobody had linked the two because nobody had written down when updates happened.
The fix was cheap. Updates moved to a documented slot, the connector setting was locked, and a test message was sent after every update. The lead also made a note on a shared calendar so two people would not update different tools on the same night.
Week three: deciding who gets woken up
The next question was practical: how urgent is urgent? Anyone can say “everything is a priority”. The lead sketched a small matrix. She combined how many customers were affected with how quickly the damage would grow, and from the two scores derived a priority number. She had read in the Atlas that a sample setup on one commercial platform derives priority from impact and urgency by adding them and subtracting one, and she copied the idea in a spreadsheet to keep it simple: a total of one meant act at once, anything higher followed a gentler clock.
Then she attached repair targets. A broken payment page would be handled the same day, any hour. A slow internal report could wait until the next working morning. Importantly, she also wrote down which hours counted, because the team did not work weekends and the old arguments had mostly been about calendar versus working time.
Week four: promises that depend on others
Several services relied on outside providers: the card gateway, the email host, an outside web developer. The lead learned from the Atlas that service levels have three layers: the promise to the customer, internal agreements between teams, and underpinning contracts with suppliers. She applied it simply. The customer promise was a same-day response to booking trouble. Internally, the coordinators agreed to triage within half an hour. Externally, the web developer’s retainer specified a four-hour response. The remaining time was held as buffer.
She noticed that the customer promise was only as strong as the weakest link, which is obvious once written down and invisible when it is not. For a company whose customers book aircraft time against hard schedules, that clarity mattered more than any new tool.
What the team changed by the end of the month
- A single running log of every disruption, with a monthly review for repeats.
- One update window, one calendar, one test message after each update.
- A simple priority formula and four repair targets with stated working hours.
- A one-page list of suppliers and their response commitments.
- An alert that tells a human when the notification mailbox has gone quiet for too long.
That last item came from another Atlas warning: if the alerting layer fails without anyone noticing, outages stay hidden. A silent inbox is exactly that kind of failure, so she set a rule to flag two hours of silence during office hours.
What she deliberately did not do
She did not buy a service-management suite. The Atlas lists heavyweight tooling imposed on a one-team, one-service operation as a common mistake, and she had no wish to join the club. She did not create an approval committee either; the pricing coordinator remained free to do routine updates in the agreed slot. She did not blame anyone. The log stuck to facts, not names.
If you run a charter business and want the same clarity on the commercial side, the plain-English pieces on how empty-leg flights operate, a beginner’s cost guide to chartering and what drives charter pricing up or down show how much of this industry rests on tidy information and dependable processes.
Lessons worth keeping
The story is invented, but its lessons track how the discipline is described in the Atlas article on ITOps. First, write things down, because memory fails exactly when you are busiest. Second, separate fixing from understanding. Third, give every promise a clock and decide which hours count. Fourth, keep changes in a visible schedule. Fifth, make sure something tells you when your monitoring stops.
None of these steps needs a budget. All of them need a person who cares enough to start the log. Often that person is already on your team, quietly restarting things on a Tuesday night, waiting for permission to do it properly.

