← ccPlanning Academy

The three building blocks of a forecast

Free visual lesson · about 5 minutes · short quiz at the end

ccPlanning academy · forecasting

The three building blocks of a forecast

Five minutes on the foundation every contact-centre forecast rests on.

The big idea

Every forecast rests on three numbers.

Get all three right on clean data and almost any sensible method gives you a usable forecast. Get one wrong and the most advanced model in the world is confidently wrong.

Building block 1

Volume

How many contacts will arrive — broken down by interval, by channel, and ideally by contact type.

It’s the most visible block, and the one that soaks up almost all the forecasting attention because it’s the one everyone can picture. But a perfect volume number, on its own, tells you nothing about how many people to roster.

Building block 2

AHT — average handle time

How long each contact takes — talk time, plus hold, plus the after-call work that finishes it off.

This is the block most often assumed when it should be forecast. AHT drifts: it moves with channel mix, with product and process complexity, and — quietly — upward as self-service skims off the easy contacts and leaves your agents a harder residual.

Building block 3

Shrinkage

The share of paid time that isn’t available to handle contacts — holidays, training, breaks, sickness, meetings, coaching, system downtime.

It’s the block most often modelled optimistically, and it’s the most dangerous place to be wrong, because it sits at the very end of the sum. A shrinkage error doesn’t average out — it flows straight through to the number of bums on seats you ask for.

How they combine

Volume × AHT, grossed up for shrinkage

Volume times AHT divided by availability gives the staffing requirement

A two-minute example

Watch the error compound

Say 1,000 calls at 5 minutes is about 83 productive agent-hours of work. Gross that up at 30% shrinkage and you need roughly 119 hours of paid time.

Now assume shrinkage is really 38%, not 30%. The same workload needs about 134 hours — 13% more staff — from one quietly optimistic input. Nothing in your volume model would ever flag it.

The trap

Nail volume, fudge the other two — still wrong.

A planner who forecasts volume to two decimal places and applies one inherited shrinkage figure all year has done the easy third of the job well…

…and the hard two-thirds badly. The requirement is only as good as the weakest of the three.

The takeaway

Forecast all three. The craft is in the fundamentals.

Volume, AHT, and shrinkage — each a thing to be forecast, not just volume.

Now test yourself ↓

1 / 8

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

In depth: why three blocks, and why the weakest one wins

Almost every forecasting conversation is really a conversation about volume. It’s the number that lands in the inbox, the line on the chart everyone points at, the thing the forecast is judged on when it misses. So it’s where the effort goes — the seasonality, the day-of-week shape, the clever model. None of that is wrong. But staffing isn’t driven by how many contacts arrive; it’s driven by how much work arrives and how much of your paid time is free to do it. Volume is only the first of three numbers that decide that, and it’s the one least likely to sink you, precisely because it gets the attention.

Volume tells you how many; AHT turns that into work

A thousand contacts could be twenty hours of work or two hundred, depending entirely on how long each one takes. That’s average handle time, and it’s the multiplier that converts a contact count into a workload. The trap is that AHT looks stable — one tidy number you can carry forward — so people freeze it and move on. In reality it drifts with the mix of contact types, creeps up as products and regulation get more complex, and shifts the moment a chatbot or self-service starts deflecting the simple stuff, because what’s left for your agents is the hard residual. Treat AHT as something to be forecast in its own right, not a constant.

Shrinkage is where small errors do the most damage

Once you know the workload, you gross it up to cover everything that takes an agent away from contacts — leave, training, breaks, sickness, meetings, coaching, the lot. That’s shrinkage, and it sits at the final step of the calculation, which is exactly what makes it dangerous. An error here doesn’t wash out against anything; it scales the whole requirement. A shrinkage assumption that’s eight points too low doesn’t make you 8% short — it can leave you double figures short of staff, every single week, with a volume forecast that looks immaculate.

The principle to remember: the staffing requirement is only ever as good as the weakest of the three blocks. Forecasting volume to two decimal places while carrying one inherited shrinkage figure all year is doing the easy third of the job brilliantly and the hard two-thirds badly. Spread the rigour across all three.

Quick quiz

Five questions. Pick an answer to each, then check your score.

1. How many building blocks does every forecast rest on?

Three: volume, AHT, and shrinkage. Miss any one and the staffing requirement is wrong.

2. Which block is most often treated as a fixed assumption when it should be forecast?

AHT. It drifts with channel mix, complexity, and self-service deflection — so it deserves its own forecast, not a frozen average.

3. Why is a shrinkage error so damaging?

Shrinkage grosses up the requirement at the final step — model it optimistically and you under-staff all year.

4. You nail the volume forecast but fudge AHT and shrinkage. Your staffing will be…

Wrong. The requirement is only as good as the weakest of the three blocks.

5. What’s the right discipline?

Forecast all three. The craft is in the fundamentals, not the sophistication of the method.