Sick Leave Spikes: Reading Them Without Guessing About People

A cluster of short sick days lands in the two weeks after a difficult release and a manager starts wondering who is really ill. That is the wrong question, and asking it out loud damages trust faster than the absences damage the schedule.

Sick leave spikes are worth reading. Read them as a team pattern with a date attached, never as a set of individual cases. The information you need sits in the timing and the count, not in anyone's reason for being off.

What a spike actually looks like

A spike is not one person taking a week. It is several short absences, from different people on the same team, inside a narrow window.

Three shapes come up often:

• Short and clustered. One or two day absences from several people within the same fortnight.
• Following a push. The window starts a few days after a launch, a deadline, or a stretch of long hours.
• Concentrated in one team. Neighbouring teams doing normal work show nothing unusual in the same period.

When all three are present, the pattern is telling you something about the fortnight, not about the people. It rarely arrives alone either, because a team past capacity leaves other quiet signs in the same weeks.

Why it follows a hard release

People push through while the deadline is live and stop afterwards. That is ordinary human behaviour and it is not a discipline problem. Anything postponed during a crunch, including rest, recovery from a minor illness and appointments, tends to arrive at once when the pressure drops.

There are other explanations. Seasonal illness moves through offices. School terms and public holidays shift patterns. A single event can put several colleagues in the same place at the same time. So treat the release as one candidate explanation among several, and check the calendar before you settle on it.

Read sick leave at team level, over a window of weeks, alongside what the team was doing. One person's absences are a private matter. A team's pattern in the fortnight after a launch is a planning signal.

How to look without prying

Set the rules before you open anything. Written rules protect the people in the data and they protect you from an accusation of snooping.

A workable set:

1. Aggregate only. Counts per team per week. No names in the view you use for this.

2. No reasons. Medical detail belongs with whoever is legally required to hold it, and nowhere else. You do not need it to see a pattern.

3. A minimum team size. In a team of three, aggregate data is not really anonymous. Set a floor below which you do not run the analysis at all.

4. A stated purpose. You are checking whether workload planning needs to change. Write that down, and do not use the same view for performance conversations.

5. A fixed cadence. Look monthly or quarterly, at the same time, for everyone. Looking only after a bad week turns analysis into suspicion.

Employment and data protection rules on absence records differ by country and by contract, so confirm what your organisation is allowed to record and who is allowed to see it before you build any of this.

What to change when you see one

The response is about the next quarter, not the last fortnight.

• Look at the run up, not the absence. How many consecutive weeks did that team spend above normal hours? If the answer is more than a couple, the schedule caused it, and a two week audit of meeting load against focus hours will show where those hours went.
• Check the recovery gap. Teams that go straight from a launch into the next sprint at full speed carry the fatigue forward. A lighter week after a heavy one is cheaper than a spike.
• Count who was load bearing. Crunch periods usually run through a small number of people, often the one quietly holding three projects together. If the same names appear in the overtime and then in the absences, spread the dependency before the next release.
• Ask the team, in a retrospective, about pace. You do not need to mention absence at all. People will tell you whether the last month was sustainable if you ask them plainly.

What not to do

Do not confront individuals with their absence counts as evidence of anything. Do not build a dashboard that ranks people by sick days. Do not use the pattern to justify tighter approval rules, which mostly teaches people to come in ill and spread it.

And do not ask a manager to interpret medical information. That is not their job and it puts them in a position where any answer they give is wrong.

The honest summary

A sick leave spike is a schedule review with a lagging indicator attached. Treat it that way and it is useful. Treat it as a list of suspects and you get a team that stops reporting absence honestly, which leaves you with worse data and the same problem.