Your Christmas range isn't "erratic" — your software just can't see seasons
Somewhere in your inventory system, your best seasonal line is probably labelled "erratic", "volatile" or "class Z — unpredictable". Because of that label, it's getting either a token buffer or a panicked, oversized one. Both are wrong, and both come from the same classification mistake that almost every inventory tool on the market makes.
How the standard grading works
Most tools sort products into classes by two letters. ABC ranks products by value — sensible enough. XYZ grades them by demand variability: X for steady, Y for wobbly, Z for wild. The Z-class products are treated as unforecastable, so the system stops trying — it either holds a huge just-in-case buffer or quietly gives up on them.
The variability measure behind XYZ is almost always the same: how much the sales history wobbles around its own average. And that's the bug.
The mistake: wobble isn't unpredictability
Think about a Christmas line. It sells 50 units a month from January to September, 400 in October, 900 in November, 1,200 in December. Measured against its yearly average, that history wobbles enormously — so the tool stamps it "Z: erratic" and effectively stops planning it properly.
But is that product actually unpredictable? It's one of the most predictable things in your warehouse. It does the same thing every year, on schedule, for the same reason. A five-year-old could forecast it: "it sells at Christmas".
Variability measures how much the sales move. Predictability measures whether you can see the moves coming. Seasonal products score terribly on the first and brilliantly on the second — and most software only measures the first.
What the mislabel costs you
- Starved buffers before the season. "Erratic" products get planned with low confidence — right up until October, when the season the system couldn't see arrives and the stockouts start. You lose your highest-margin weeks of the year to a labelling error.
- Panic orders at the worst prices. The mid-season emergency top-up comes with air freight, broken price breaks and whatever stock the supplier has left.
- January overhang. The panicked over-order lands late, the season ends, and the surplus joins your dead stock pile — where it gets blamed on "unpredictable demand" rather than on the classification that caused it.
The fix: grade products on forecastability
The right question isn't "how much does this product's history wobble?" — it's "when we actually forecast this product, how wrong are we?"
Run a proper forecast on the seasonal line — one that can learn a yearly rhythm — and test it against the product's own past: hide the last year of history, predict it, compare. A genuinely seasonal product comes out with small forecast errors, because the pattern repeats. A genuinely erratic product — the one whose spikes have no rhythm at all — comes out with big errors no matter what the forecast tries. That is the honest dividing line between "plan with confidence" and "hold extra insurance".
Grade products by forecast error instead of historical wobble and the classes rearrange dramatically: Christmas lines, summer lines, back-to-school lines move from "erratic" to "predictable — with a season". Buffers land before the peak, sized for the peak. The genuinely wild products are still flagged — but now the label is earned, not guessed. This is exactly how Optimal Chain grades every product: it runs the forecast tournament first, measures each product's real forecast accuracy, and classifies from that — so a product is only ever called unpredictable after the maths has genuinely tried to predict it.
What you can check this week
- Pull your Z-class (or "erratic") list and scan it for products you personally know are seasonal. Every one you find is a mislabel costing you either sales or cash.
- Look at last year's stockouts by month. A cluster in October–December is the classic signature of seasonal lines being planned as erratic.
- Check where your dead stock was born. If a chunk of it is seasonal surplus ordered mid-season at panic prices, the label — not the demand — was the root cause.
See your own catalogue graded properly
Send us a sales export — any system, Excel included — and the free blueprint will show you every product graded by real forecastability, with the seasonal lines identified and their buffers timed to the season. It's often the single biggest surprise in the diagnosis. No card, no sales calls — and if you want the full remedy afterwards, the price list is published.