A Field Guide to Rank Chart Shapes

Rank series produce a small vocabulary of shapes. Recognising them, and knowing which mundane cause produces each, resolves most charts in seconds.

Sawtooth

Appearance: sharp improvements, gradual decay, repeated.

Usually means: individual sales are separable. A low-velocity product.

Useful property: counting teeth over a period approximates the sales count, which is a more direct estimate than any conversion table at this velocity.

Rule out: nothing. This is the normal appearance of a slow product.

Smooth line

Appearance: low variation, gradual movement.

Usually means: sales frequent enough to overlap. Higher velocity.

Rule out: an over-smoothed chart. Check whether a moving average is being displayed as raw data.

Spike with decay

Appearance: sharp improvement, then a return toward the prior level over days.

Usually means: an event. Promotion, publicity, price cut, feature placement.

Rule out: a competitor stock-out that temporarily removed a product above you.

Step change

Appearance: a move to a new level that persists.

Usually means: something structural. Price change, listing change, category change, or a genuine demand shift.

Rule out first: a collection method change or a category restructure. Check the method log before the market.

Decay and sharp recovery

Appearance: smooth decline over days, then a rapid return to the previous level.

Almost always means: a stock-out and a restock.

Rule out: nothing else produces this shape. Genuine demand decline does not rebound cleanly.

This is the single most useful shape to recognise, because it prevents the most common false conclusion in competitive analysis.

Slow drift downward

Appearance: gradual worsening over weeks with no events.

Usually means: declining velocity, or a growing category pushing everyone down.

Rule out: the basket. If comparable products drifted too, the category grew and nothing happened to this product.

Synchronised movement across many products

Appearance: several tracked products moving together.

Usually means: a market event, a seasonal transition, or a category restructure.

Rule out: a collection problem affecting the whole run. Check the health metrics before the market.

Vertical discontinuity

Appearance: an instantaneous jump with no transition.

Almost always a data problem. A changed identifier, a switched rank type, a parser reading a different element, a merged listing.

Rule out in this order: identifier, rank type, method change, category restructure. Only then consider a market cause, and a genuine market event essentially never produces a truly vertical move in a series with reasonable cadence.

The order of investigation

For any shape that looks like news:

Is the data intact? Gaps, method changes, identifier changes.

Was it available? Stock-outs explain more than anything else.

Did the price change?

Did the basket move?

Is it seasonal?

Only then, is it about the product?

Five checks, a few minutes, and they resolve the large majority of charts that would otherwise become a slide.

Building a shape library from your own data

The general patterns are a starting point; your category has its own vocabulary.

Collect examples as you encounter them. When a shape is explained, save the chart with the explanation.

Label each with the cause, confirmed rather than assumed.

After a year you have a reference specific to your category and your products, which is far more useful than any general guide.

It also trains new analysts fast. Handing someone twenty annotated charts with their causes teaches pattern recognition in an hour that would otherwise take a year of exposure.

Include the ambiguous cases. Shapes that were never satisfactorily explained belong in the library too, labelled as unexplained, because the next occurrence may be the one that resolves them.