Rank your catalogue by revenue and you get one ordering. Rank it by money kept and you get another. The products that move between the two lists are where the decisions have been going wrong — the celebrated best seller that funds nothing, and the quiet SKU in fourth place that pays for the warehouse.

ORVX computes the second ordering.

Profit per SKU, after everything

Per product Realized revenue delivered units × price paid − COGS delivered units × landed unit cost = Gross profit − Allocated ad spend the campaigns that sold this product = Contribution after ads

Three things in that calculation are routinely got wrong, and each of them changes the ranking:

  • Units, not orders. Revenue counts delivered units at the price actually paid, so a discounted unit and a full-price unit are not treated as the same sale.
  • Variant-level cost. Where a variant has its own landed cost, it is used. A catalogue-level average silently moves profit from one size to another and makes the extremes of a size run look like the middle.
  • Returns. Returned units never enter revenue, so a product with a 30% return rate does not get to keep the gross profit on the 30%.

Collections, not just products

Merchandising decisions are usually made at the collection level, so the same arithmetic runs there: realized revenue, COGS, gross profit and margin, ad spend, contribution after ads, unit return rate and remaining inventory — per collection.

Two figures at that level tend to be the useful ones. Revenue share tells you how much of the business a collection carries. Product concentration tells you how much of that collection is really one SKU — which is the difference between a healthy category and a single product with a category built around it.

The ad allocation problem, stated honestly

Attributing ad spend to individual products is the part of SKU profitability that most analyses quietly fudge. Campaigns are rarely one-product, platforms disagree with each other about attribution, and any allocation rule is a judgement rather than a fact.

ORVX allocates spend where the link between campaign and product is recorded, and reports contribution after ads only where that link exists. Where it does not, you get gross profit — a true number — instead of a contribution figure resting on an invented split. A product-level ROAS that looks precise because the software picked a rule for you is worse than an honest gap.

What the ranking usually reveals

  • A top-three seller whose contribution after ads is near zero, because it is bought almost entirely through paid traffic at a margin that cannot carry the CPA.
  • A product with an unremarkable revenue line and the best margin in the catalogue, which has never had budget because nobody looked past the sales report.
  • A collection that appears healthy in aggregate and turns out to be one SKU subsidising four that lose money.
  • Products with no cost recorded, flagged rather than shown at 100% margin.

The method, if you want it first

The arithmetic here is not proprietary and is worth understanding before trusting any tool with it. The companion guide walks through the same calculation by hand, including the parts where it is easy to mislead yourself.