The cost of your service level

Deep-dive lesson · about 10 minutes · short quiz at the end

ccPlanning academy · metrics · deep dive

The cost of your service level

80/20 isn’t a law of nature. It’s a number you should be able to justify.

The problem

Where did 80/20 even come from?

“80% of calls answered in 20 seconds” is quoted like physics. It isn’t. It has no scientific basis — it’s an old telco convention that stuck.

Which means it might be right for you, or far too tight, or far too loose. You can’t know until you cost it.

One side of the scale

Faster answers cost agents.

To answer more calls more quickly, you need more people waiting and ready. Every step up in service level adds staff — and staff are 60–70% of operating cost.

Push the target higher and the labour bill climbs with it, faster and faster near the top.

The other side

Slower answers cost customers.

Answer too slowly and callers abandon. Some call back; some don’t; some were about to buy. Each lost contact has a value — a sale, a retained customer, a complaint avoided.

Run the service level too low and you save on staff while quietly bleeding that value away.

The trade-off

Two costs, moving in opposite directions.

As service level rises, labour cost goes up and the cost of lost contacts goes down. Add them together and the total has a shape: a valley.

Labour rises, lost value falls, and the total is lowest somewhere in the middle.

The answer

The right target is the bottom of the valley.

The cost-optimal service level is the one where total cost — labour plus lost value — is at its minimum. Not the highest service you can afford, and not the cheapest staffing; the point where the two balance.

That point is derivable. It just needs the value of a contact.

It depends

A sales line and a cost centre have different optima.

If each contact is worth a £200 sale, slow answers are expensive — the optimum sits high. If contacts are low-value queries, the same abandons cost little — the optimum sits lower.

One service-level target across every queue almost guarantees you’re over-serving some and under-serving others.

Near the ceiling

The last few points cost the most.

Going from 80% to 90% costs some agents. Going from 95% to 99% costs a lot — you’re paying for people who sit idle most of the time, just in case. Diminishing returns are brutal at the top.

That’s usually well past the bottom of the cost valley — you’re buying service nobody values at a price that hurts.

Make it concrete

Sweep the staffing, find the floor.

Take an interval, try every agent count, and for each one add the labour cost to the value of the contacts abandoned. Plot the total and the cheapest point jumps out — often a service level you’d never have picked by tradition.

The cost-optimal service level calculator does exactly this, and draws you the valley.

The takeaway

Derive the target. Don’t inherit it.

80/20 is a starting guess, not an answer. Cost the labour against the value of lost contacts, find the bottom of the curve, and set a target you can defend in pounds — per queue, because they’re not all the same.

Now test yourself ↓

1 / 10

Slides done? Here’s the same idea in a bit more depth — the part worth keeping.

In depth: the service level is a cost decision, not a tradition

The 80/20 service-level target — 80% of contacts answered within 20 seconds — is one of the most-quoted numbers in the contact centre, and one of the least examined. It has no scientific basis; it’s a convention inherited from early telephony that hardened into a default. That doesn’t make it wrong, but it does make it a starting guess rather than an answer. The only way to know the right target for a given queue is to treat the service level as what it actually is: a balance between two costs that move in opposite directions.

The two costs, and the valley between them

Raising the service level means answering faster, which means more agents waiting and ready — so labour cost rises as the target climbs, and rises fastest near the top, where you’re paying for staff who are idle most of the time. Lowering the service level saves that staff cost but lets more callers abandon, and every abandoned contact carries a value: a lost sale, a customer who churns, a complaint that escalates. As the target falls, that lost value grows. Add the two together and the total cost forms a U — high at both ends, lowest somewhere in the middle. That bottom point is the cost-optimal service level, and it’s the target you can actually defend.

Why it differs by queue, and what to do

The location of the valley depends almost entirely on the value of a contact. A sales line where each call might be a £200 order has an expensive downside to slow answers, so its optimum sits high; a low-value query line has a cheap downside, so its optimum sits lower. Running one blanket target across every queue therefore over-serves some and under-serves others — spending money where it isn’t valued and skimping where it is. The practical method is to sweep the staffing for each queue, add labour cost to the value of abandoned contacts at every level, and read off the minimum. More often than not it lands somewhere other than 80/20 — which is exactly the point.

The principle to remember: the service level is a cost decision. Find where labour cost plus the value of lost contacts is lowest, set that as the target, and do it per queue — don’t inherit 80/20 and hope.

The cost of your service level