Planning benchmarks — what’s “normal”?
The numbers planners get asked about most — “is our shrinkage too high? is 85% occupancy okay? what attrition is normal?” Here are the typical ranges and what sitting outside them tends to mean.
| Metric | Typical range | What it usually means at the edges |
|---|---|---|
| Shrinkage | ~28–35% | Above ~40% usually means hidden off-phone time nobody is counting; below ~25% often means under-investment in training, coaching or breaks. |
| Occupancy (voice) | ~80–87% healthy | 87–92% is a caution band; sustained 92%+ is a burnout warning that feeds attrition, not a sign of efficiency. |
| Annual attrition | ~20–40% (UK voice) | Hugely variable; under ~20% is strong for high-volume voice, over ~50% is usually a culture, pay or management problem, not a planning one. |
| Service level | 80% in 20s (convention) | The famous “80/20” has no scientific basis — it’s a convention. Set the target to what your customers tolerate and what the contact is worth. |
| Abandonment | ~2–5% (often targeted) | Driven by how long callers will wait; a low rate with long waits can hide short-abandons. Outbound predictive diallers are legally capped (UK: 3%). |
| Adherence | ~85–90% (commonly set) | Conformance to the plan, not maximum login time. Chasing it much higher buys gaming, not coverage. |
| Forecast accuracy | ~5–15% WAPE (interval) | Smaller, noisier operations sit higher; the figure that matters most isn’t the size of the error but its bias — a forecast that’s always short hurts far more. |
| Chat concurrency | ~2–3 (complex chat) | Capacity rises with concurrency but less than proportionally; push it too high and response times, quality and retention all sag. |
| AHT | no benchmark | AHT varies so much by channel, sector and contact type that cross-operation comparison is meaningless. Watch your own trend and mix, not someone else’s number. |
The detail behind the ranges
Shrinkage — ~28–35%, but measure your own
A healthy operation that counts leave, training, coaching, breaks, meetings and unplanned absence honestly tends to land around 28–35%. The number is almost meaningless to compare between operations, though, because definitions vary wildly — the value is in measuring your own from actuals and forecasting it by period.
Deeper: Shrinkage: the planner’s hardest input · Shrinkage, Honestly (white paper)
Occupancy — healthy 80–87%, danger above the low-90s
Occupancy is an output of staffing and your service target, not a lever to set, and it rises with size. A small team hitting service might run at 80%; a large one at 90% on the same target. Sustained occupancy in the nineties is a burnout warning that feeds the attrition spiral, not a badge of efficiency.
Deeper: Occupancy: the metric that burns out your agents · The Occupancy Trap (white paper)
Attrition — ~20–40% for UK voice, but it’s contextual
Annual attrition spans an enormous range by sector and labour market. For high-volume UK voice operations, 20–40% is common; outsourced and entry-level operations run higher. What matters for planning is less the headline rate than its shape — early-life attrition, seasonal spikes, the post-bonus exodus — because that’s what you have to forecast and hire against.
Deeper: What ‘normal’ attrition actually is · The true cost of attrition
Service level — 80/20 is a habit, not a law
“80% answered in 20 seconds” is the most quoted target in the industry and has no scientific basis whatsoever. It’s a convention that many centres adopt unthinkingly and then over- or under-invest against. The right target depends on what your customers tolerate and what each contact is worth.
Deeper: Setting the right service level target · Is service level a dead KPI?
The honest caveat
If a vendor or consultant hands you a precise “industry benchmark,” ask how it was measured and whether it counts the same things you do. Most don’t. The most useful benchmark is almost always your own operation last year, measured consistently — that’s the comparison that actually tells you whether you’re improving.
Want to put numbers on your own operation? Try the shrinkage, cost-of-attrition and Erlang C calculators, or work through the Academy.