Similar names can hide different denominators

The most common commerce analytics error is not bad arithmetic. It is discussing metrics with different denominators as if they were interchangeable.

GMV, units, orders, buyers, average order value, and average selling price all describe commercial activity, but they answer different questions. Before using any metric, identify its numerator, denominator, time window, order-state rules, and aggregation level.

Without those definitions, the same label can produce several valid but incompatible numbers.

The three metrics in AsKlear’s current contract

In the current JD.com monthly data contract:

The denominator of ASP is item volume. It answers “How much sales value did the average item contribute?” It does not answer “How much did the average order spend?” or “How much did the average buyer spend?”

Average order value requires order count. Spend per buyer requires a distinct buyer count. If those fields are unavailable, ASP must not be used as a substitute.

  • GMV is aggregated post-coupon sales value for the selected scope.
  • Units is aggregated item sales volume for the selected scope.
  • ASP is average selling price, calculated as GMV divided by units when units are not zero.

Use the first decomposition

Within one consistent query scope, begin with a simple identity:

GMV = Units × ASP

This separates sales-value movement into volume and average selling price. It is not a complete operating funnel, but it is highly useful for monthly product sales analysis.

These combinations are diagnostic starting points, not proof of cause. Promotions, stockouts, launches, and assortment shifts still require validation.

  • GMV and units both rise: broader scale growth is likely.
  • GMV rises while units stay flat: ASP or premium product mix increased.
  • Units rise while GMV stays flat: ASP fell, possibly through discounting or lower-priced mix.
  • GMV and units both fall: investigate demand, availability, and assortment changes.

Do not confuse a product identity with a funnel

Popular analytics frameworks often decompose GMV further into traffic, conversion, and average order value. That is appropriate when exposure, click, order, and buyer data are available.

A monthly product sales dataset may not contain those funnel fields. If the available evidence is GMV, units, and ASP, the analysis should stop at the level the data can support. It cannot reverse-engineer traffic or conversion.

Good analysis is not the longest possible formula. It clearly separates what the current dataset proves from what requires another source.

Aggregated ASP is weighted

Brand- or category-level ASP is not the simple average of product prices. It is total GMV divided by total units, so high-volume products carry more weight.

A brand can report a higher ASP even when no individual product changes price. If more volume shifts toward premium products, the aggregate rises because the mix changed.

When ASP moves materially, check at least two things: whether the unit share of leading products changed, and whether new products entered or old products left the scope.

Compare with the same ruler

For month, brand, or category comparisons, keep these conditions fixed:

Read absolute values alongside rates of change. Moving from 1 to 3 produces a large percentage increase but may remain commercially small. Five percent growth on a large base can create much more absolute value.

  • The same complete-month window.
  • The same category level and filters.
  • The same entity-resolution rules.
  • The same metric definition and aggregation.

A practical reading order

Read a result in this sequence:

If the conclusion concerns share, add the matching category total as a denominator. If it concerns margin, traffic, conversion, or repeat purchase, bring in additional data rather than inferring those outcomes from these three metrics.

The goal is not to collect more metric names. It is to explain every number precisely—and to know what it cannot explain.

  • GMV first, to establish scale and direction.
  • Units next, to test whether volume drove the movement.
  • ASP next, to identify price or mix movement.
  • Product, brand, or category detail last, to locate where the change concentrated.