Call centre shrinkage is the share of paid time your agents are not available to take calls, and it is the reason 25 agents on an Erlang sheet becomes 36 people on the rota. If you skip it, you will staff to the sheet and miss your service level every single week, and nothing in the sheet will tell you why.
In my post on building an Erlang C calculator in Excel, the worked example landed on 25 agents for 120 calls in a half hour. I mentioned shrinkage at the end and moved on. It deserves more than that, because it is where the clean maths meets the actual building.
What counts as call centre shrinkage?
Anything that takes a paid person off the phones when the forecast says calls are coming. The Erlang calculation gives you agents logged in and available. Shrinkage is everything between that number and the people on your payroll.
There are two kinds, and the difference matters when you plan.
- Planned shrinkage. Breaks and lunches, team meetings, coaching sessions, classroom training, annual leave, public holidays. You know about these in advance and you can schedule around them.
- Unplanned shrinkage. Sickness, lateness, a system outage that stops people working, a fire drill, the agent who is stuck in a long call at the end of their shift. You can only forecast these from history.
Most teams track the first kind and guess at the second. Guessing is fine for a month. After that, pull the numbers.
How do you calculate shrinkage?
The formula is one line. Take the hours your people were paid for, subtract the hours they were actually available to take contacts, and divide by the paid hours.
Here is an example with made-up but realistic figures for one week. Treat it as a template for your own, not as a benchmark:
- Breaks and lunches: 10% of paid time
- Meetings and one-to-ones: 3%
- Training and coaching: 4%
- Annual leave and public holidays: 10%
- Sickness and lateness: 3%
Add them up and you get 30%. That means for every 100 paid hours, 70 are available for calls. To turn a required agent count into a rostered headcount, divide by 0.70, not by 0.30 and not by subtracting 30%. The Erlang sheet said 25. So 25 divided by 0.70 is 35.7, which you round up to 36.
If you took the shortcut and added 30% to 25, you would get 32.5. That is three or four people short, in every half hour, all day. It is the most common arithmetic mistake in a first staffing sheet, and nothing on the sheet warns you about it.
What happens when shrinkage moves?
A few points of shrinkage sound small. Run them through the same arithmetic and they are not.
- At 25% shrinkage, 25 agents needs 33.3 people, so 34 on the rota.
- At 30%, it needs 35.7, so 36.
- At 35%, it needs 38.5, so 39.
- At 40%, it needs 41.7, so 42.
Going from 25% to 35% adds five people to cover the same 120 calls. Nothing about the customers changed. The phones are not busier. You simply have fewer hands available because more of your week is going on leave, training and meetings.
That is why I do not trust any staffing plan that quotes one shrinkage figure for the whole year. August leave looks nothing like October. A month with a product launch and a week of classroom training for new starters looks nothing like a quiet February.
Where do the hours actually go?
Take the example above and a team of 36. At 30% shrinkage, about eleven of them are off the phones at any given moment. Nobody is slacking. That is simply what a normal week looks like when you add up lunches, a team meeting, someone on holiday, someone in training, and one person home with a cold.
The uncomfortable part is that the biggest line is often annual leave, and it is the one that managers are most reluctant to plan around. A manager who approves four people off in the same week, because they all asked nicely, has just taken the Erlang sheet out of the building. The calculator cannot protect you from that. Your leave policy can.
Training is the other big one. In 2007, Microsoft shipped Office 2007 with the new Excel file format, and plenty of teams spent days getting agents and analysts used to a new ribbon interface. Any change like that, whether it is a new CRM, a new Excel, or a new call script, shows up as shrinkage the week it lands. Put it on the plan before it happens.

How do you keep shrinkage honest?
Three habits will take you most of the way there.
- Measure it by interval, not by month. Shrinkage is lumpy. Breaks cluster around midday, meetings land on Tuesday mornings, and leave bunches around school holidays. A monthly average hides the half hours where you are actually exposed.
- Split planned from unplanned. Planned shrinkage goes on the rota in advance. Unplanned shrinkage goes into the headcount as a buffer, based on the last twelve weeks of sickness and lateness, not on hope.
- Check the sheet against the result. If the rota says 36 and the phone system shows 31 people actually logged in at 10am on a Monday, the gap is your real shrinkage. Compare it with what you assumed.
I thought shrinkage was an admin detail you tidied up after the real planning was done. It is the planning. The Erlang formula tells you what the calls need. Shrinkage tells you what the building has to supply to deliver it.
Does shrinkage hit every half hour equally?
No, and this is where a single percentage lets you down. Go back to the two intervals from the calculator post. The busy half hour of 120 calls needs 25 agents, and at 30% shrinkage that is 36 people. The quiet half hour of 60 calls needs 14 agents, and 14 divided by 0.70 is 20 people.
So the same team has to flex between 36 and 20 across the day. That is a sixteen-person swing, and it is the reason the lunch rota matters so much. If twelve of your 36 go to lunch at 12:00 because that is when everyone has always gone, you are 12 people down in what might be your busiest hour. Stagger the breaks against the forecast, not against habit.
The good news is that the swing works in your favour as well. The quiet intervals are where meetings, coaching and training belong, because 20 people can cover them. The planning job is to push planned shrinkage into the troughs and keep the peaks clear. A rota that does that can run at 30% shrinkage and still hit 80/20. A rota that puts the team meeting at 10am on Monday cannot.
Should you ever cut shrinkage?
Sometimes, but be careful what you cut. Breaks and lunches are not negotiable, and squeezing them leads to sickness and attrition that cost you far more than the minutes you saved. Meetings are the easiest to trim: move the weekly team meeting to the quietest half hour of the week and you recover people for peak without losing the meeting.
Training is more subtle. Putting it off protects this week’s service level and hurts next quarter’s. The better move is to schedule it into the troughs on purpose, which means you need the Erlang sheet filled in for the whole day, not just for the busiest half hour. If you have not done that yet, the Wikipedia entry on the Erlang unit is a quick reminder of what the formula is actually measuring, and Call Centre Helper’s introduction walks through the same logic from the other side.
What is the quickest way to start?
Pull last month’s timesheets and your phone system’s logged-in hours. Work out available hours divided by paid hours for each week. That number, one week at a time, is your shrinkage. It will probably be higher than the figure in your head.
Then rebuild the rota from the sheet using the real figure. If the Erlang sheet from my first post in this series says 25 and shrinkage says 36, put 36 names on the rota and see what the gaps look like. Compare that with your current rota. The distance between the two is the service level you have been losing without knowing it.
The sheet was never wrong. It was answering a smaller question than the one you were asking.
Hutch Morzaria is a Director-level CX and Support Leadership professional with 19 years of experience building global support organizations across SaaS, Fintech, and enterprise technology. He has hired dozens of support and CX leaders across his career and holds ITIL Expert certification across V3 and V4.



