More contact centre leaders are planning to shrink the agent team in 2027, and most of them plan to do it the quiet way: stop hiring and let attrition do the work. A Gartner survey cited by No Jitter found that 63% of customer service leaders are reducing agent headcount through attrition, and another 31% are planning or running layoffs. It sounds clean. No exits to manage, no cost of severance.
For a workforce planner the question is different: does attrition remove the right people, in the right places, at the right speed? It rarely does. This guide uses a simple model of a 200 agent operation to show where attrition-only reduction goes wrong, and what to plan instead.
Why attrition feels like a safe plan
Contact centre attrition is high. The same No Jitter roundup reports a Verint finding that 31% of agents plan to leave within six months, and attrition in the 30% to 45% range is common in many operations. If 35% of a team leaves in a year anyway, a hiring freeze looks like a free headcount cut.
The trouble is that attrition has three features that make it a poor steering wheel:
- It does not stop when you reach the target. It keeps running at the same rate.
- It does not follow demand. People leave when their life or job market says so, not when your volume falls.
- It is random by team. Leavers come from every shift, skill and client, in uneven lumps.
The setup
We modelled one operation with these assumptions:
- 200 agents today, with headcount matched to demand.
- Demand falls by 20% over 12 months, evenly, because AI and self service take simple contacts. The team needs about 160 agents by month 12.
- Attrition of 25%, 35% or 45% a year, applied evenly each month.
- No hiring delay or ramp time. This makes the results kinder than real life.
Problem 1: attrition alone cuts too deep
Even at 25% attrition, a full freeze overshoots a 20% demand cut and leaves the team about 10 agents short. At 35% it is about 30 short, and at 45% about 50. The freeze has to end early, and you have to start hiring again just to stop the team shrinking below demand.
So the real plan is not "no hiring". It is hiring at a lower, steady rate, and the number depends on both the size of the demand cut and the attrition rate.
How many hires does the plan need?
This table shows the hires needed over 12 months to keep headcount in line with falling demand (model output, rounded).
| Demand falls by | At 25% attrition | At 35% attrition | At 45% attrition |
|---|---|---|---|
| 10% | 30 to 35 hires | 60 to 65 hires | 90 to 95 hires |
| 20% | 10 to 15 hires | 35 to 40 hires | 65 to 70 hires |
| 30% | 0 hires | 10 to 15 hires | 40 to 45 hires |
Read it from the bottom left. Only a large demand fall combined with low attrition lets you stop hiring completely. In most real cases the team still needs a steady stream of hires, often 15% to 30% of the current headcount over the year. Teams with high attrition still need a large hiring engine even while shrinking, which is a cost the "freeze" story hides. The hidden capacity cost of attrition explains why each of those hires also costs weeks of reduced output.
Problem 2: leavers do not come from where you need them to
The numbers above work on the total. Your schedule does not run on the total. It runs by shift, skill, language and client. To test this we split the 200 agents into teams and let leavers fall at random. Total headcount was topped up when it fell below total demand, but the new hires landed at random across teams, which is what a general "backfill when we can" approach does.
| Team size | Months with at least one team more than 10% short | Share of teams more than 10% short in a typical month |
|---|---|---|
| 10 agents | about 9 months in 10 | about 1 in 4 |
| 25 agents | about 7 months in 10 | about 1 in 7 |
| 50 agents | about 2 months in 10 | about 1 in 16 |
At 35% attrition, an operation split into 10 agent teams (a language queue, a night shift, a niche client) has a team that is badly short almost every month, even though the total headcount looks right. Smaller pools are hit hardest because one or two exits is already 10% to 20% of the team. The same maths explains why pooling skills helps so much: a bigger pool absorbs random exits.
When the same model backfilled each team to its own need, no team fell short, but that assumes instant hiring. In real life the gap between a person leaving and a trained replacement reaching the floor is months, so the true picture sits between the two.
What to plan instead
The No Jitter data also carries a warning: 58% of agents say they are stressed at least half the time (Omdia), and agents are being pushed into constant escalation work as bots take the simple contacts. A team that is quietly shrinking while handling harder work is the team most likely to speed up its own attrition. Plan with that in mind.
- Set the target headcount by month, not by year. Take the demand forecast, apply productivity and shrinkage, and build a month by month headcount need. Compare it with the expected path of attrition.
- Plan hires as a flow. Use the table above with your own attrition. A monthly hiring rate that tracks the gap is steadier than a freeze followed by an emergency class.
- Manage by team, not just by total. Report the headcount gap for every shift, skill and client. Backfill the teams that are short, and let others shrink. Use the absence buffer calculator to see how much extra cover a small team needs on any given day.
- Move people before you hire. Cross train and re-skill agents from shrinking queues into growing ones. This keeps experienced people and avoids new ramp cost. See why a hiring freeze is not a staffing plan.
- Set a review trigger. If volume falls slower than planned, you will be short. If it falls faster, you will be long. Agree in advance what number of agents short or long moves you from hiring to freezing, and back.
- Check the AI assumption. If the demand cut depends on AI deflection, test it against live data before you stop hiring. Deflection often leaves longer, harder contacts behind, as covered in why AI deflection breaks your staffing model.
What to take to your leader
- The demand path by month and the headcount it needs.
- The attrition you expect, and the hires needed to keep up with it. A "no hiring" plan is rarely possible unless attrition is low or the demand cut is large.
- The teams that go short first, and the cost of covering them with overtime or overflow.
- The trigger that changes the plan.
Key takeaways
- A hiring freeze overshoots most demand cuts. In the example, a 20% fall in demand with 35% attrition leaves the team about 30 agents short after a year.
- Most plans still need a steady flow of hires, often 15% to 30% of headcount over the year, even while the team shrinks.
- Attrition is random by team. With 10 agent teams, some team is badly short in most months even when the total looks fine.
- Plan month by month and team by team, move people between queues first, and agree the triggers for restarting or stopping hiring.
The numbers in this article are model outputs based on the stated assumptions: even demand decline, even monthly attrition, no hiring delay. Use your own attrition, ramp and demand data for planning. Survey figures are from [No Jitter](https://nojitter.com/contact-centers/7-stats-suggest-decline-in-human-cx-agent-employment) (July 2026), quoting Gartner, Verint and Omdia; trend context from [Peopleware](https://blog.peopleware.com/7-important-workforce-management-trends-you-need-to-know-about-in-2026).