Reading a volume curve

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

ccPlanning academy · forecasting

Reading a volume curve

Train your eye to see the four patterns hiding in a single line.

The big idea

One line, four patterns.

A volume history looks like one wiggly line. It’s really four patterns layered on top of each other — and good forecasting starts by pulling them apart.

Pattern 1 of 4

Trend

The slow drift over months and years — a growing customer base, a shrinking one, channels migrating to digital.

Miss the trend and you under- or over-staff a little more every month.

Pattern 2 of 4

Seasonality

The repeating within-year shape — the January spike, the summer lull, the pre-Christmas wall.

Repeats on a yearly cycle — so last year is your best guide, adjusted for trend.

Pattern 3 of 4

Day-of-week

The shape within a week. Monday is rarely Saturday. For most operations the week has a fingerprint that barely changes.

Pattern 4 of 4

Intraday

The shape within a day — the mid-morning peak, the lunch dip, the afternoon hump. This is the one that actually drives interval-level staffing.

We go deep on this one in the next lesson.

They’re not separate

The patterns stack — and they multiply

A Monday in your January peak isn’t “a bit busy.” It’s the trend × the seasonal high × the Monday uplift × the mid-morning intraday spike, all landing at once.

That’s why a flat “plus 10%” never works: the layers compound, so the same busy interval can be wildly busier on the worst-case stack than your daily average suggests.

The takeaway

Decompose before you forecast.

Trend, seasonality, day-of-week, intraday. Name the patterns that are present, and your method has to handle each one you find — not just the wiggle.

Now test yourself ↓

1 / 7

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

In depth: why a curve is never just one thing

When a volume history is plotted as a single line, the eye sees chaos — a jagged thing that jumps around with no obvious rule. The skill of a forecaster is to stop seeing one line and start seeing the four separate, well-behaved patterns stacked inside it. Each one moves on its own clock, each has a different cause, and each needs handling differently. Pull them apart and the chaos resolves into something you can actually predict.

The four clocks

Trend is the slowest hand — the drift over months and years as your customer base grows or shrinks, or as contacts migrate from phone to digital. Seasonality is the repeating within-year shape: the January spike, the summer lull, the pre-Christmas wall, the same broad story every year. Day-of-week is the fingerprint inside a week; for most operations Monday is reliably the heaviest day and the weekend the lightest, and that shape barely changes. Intraday is the fastest hand — the shape inside a single day, the mid-morning peak and lunchtime dip — and it’s the one that actually decides how many people you need in each half-hour.

Why it matters that they compound

The patterns don’t add, they multiply. Your busiest interval of the year is the moment all four hands point the same way at once — an upward trend, the seasonal peak, the busiest weekday and the daily spike, stacked. That’s why a flat percentage uplift quietly fails: it spreads evenly across a curve whose peaks are driven by layers piling up, so it under-staffs the moments that matter most and over-staffs the quiet ones. Naming which patterns are present also tells you which method you need — a simple average can cope with day-of-week, but it can’t see a trend or a moving seasonal peak.

The principle to remember: decompose before you forecast. Name the patterns that are actually present in your data, and make sure your method handles every one you find — not just the surface wiggle.

Reading a volume curve