Category Ranks Versus the Overall Rank

Most marketplace products display more than one rank: a position in the whole catalogue and positions within one or more categories. They are computed from the same sales but they are not interchangeable.

The overall rank

Position across the entire catalogue, which may contain tens of millions of items.

Best for: tracking a single product's trajectory over time, and comparing products in different categories where a category comparison is impossible.

Weak for: understanding competitive position, because the field includes everything.

Highly compressed at the bottom. Below a few hundred thousand, movements are dominated by noise.

Category ranks

Position within a defined subset, and a product typically sits in several categories at once, at different depths of a hierarchy.

Best for: competitive analysis, because the comparison set is meaningful.

More sensitive. A small sales change moves category rank more than overall rank, which makes it a better early signal and a noisier one.

Category definitions are set by the marketplace and by the seller's classification choice, which introduces a problem covered below.

The narrow category problem

Products are frequently listed in a very specific sub-category with few competitors.

A product can be "number one" in a category containing eleven items. This is technically true and it is the basis of a great deal of marketing that means nothing.

Deep sub-categories produce impressive ranks for modest sales, which is exactly why sellers choose them.

When evaluating a competitor's claim, look at the category size and depth. A bestseller badge in a fourth-level category is not comparable to one in a top-level category.

When tracking, use a category at a level where the comparison set is genuinely comparable — usually one or two levels down, not five.

Category selection as a tactic

Sellers choose which categories to list in, within the marketplace's rules.

The incentive is obvious: a less competitive category produces a better rank, a badge, and better visibility in category browsing.

Marketplaces have rules against misclassification and enforcement is inconsistent.

For analysis this means category membership is a choice, not a fact. Two genuinely competing products may sit in different categories and never appear in each other's rank lists.

Build your comparison set by product, not by category. Decide which products actually compete, then track those, regardless of how they are classified.

Which to track

Overall rank for a long-run view of a single product, and for anything that needs to be comparable across categories.

A chosen category rank for competitive tracking, at a level you selected deliberately.

Both, recorded, because they diverge and the divergence is informative: a product improving in category while flat overall means its category is shrinking or its competitors are weakening.

Mixing them, which happens constantly

Do not compare a category rank to an overall rank. Different scales, different fields, no relationship.

Do not compare category ranks across different categories. Rank 50 in a category of 300 and rank 50 in a category of 300,000 are unrelated.

Do not apply a rank-to-sales table to a category rank unless the table was built for that category rank specifically. Almost all published tables are for overall rank.

Record which rank you captured, in the data. A column labelled "rank" with a mixture of category and overall values is a dataset that cannot be fixed later, and this happens to people who added category tracking partway through a project.

Changes to the category tree

Marketplaces restructure their category hierarchies periodically, without notice.

Categories are merged, split, renamed and reparented. A rank series spanning such a change is discontinuous and the discontinuity looks like a market event.

Record the category identifier and name at capture time, so that a break can be identified afterwards.

Check for structural changes before interpreting any large simultaneous shift affecting many products in one category. The market rarely moves that uniformly; the taxonomy does.

Recording the category properly

Category rank without the category recorded is unusable, and the recording needs more than a name.

The full category path, not just the leaf. "Home > Kitchen > Small Appliances > Coffee > Espresso Machines" carries information that "Espresso Machines" does not.

The category identifier where the marketplace exposes one, because names change and identifiers are more stable.

The reported product count where available, which settles depth questions immediately.

The capture date, because hierarchies are restructured.

All four, on every observation. They cost nothing to store and they are what allows a discontinuity to be diagnosed as a taxonomy change rather than a market event a year later, when nobody remembers.

When a large simultaneous shift affects many products in one category, check the recorded paths before the market. Categories are restructured more often than markets move that uniformly.