What does the product analyst show?
For any range, every product and variant with its units, its net sales after realised discount, its returns and what they cost, its cost, and the profit left after all of that. The lines are sorted by money, so the top of the winners list is the product that earned the most and the top of the losers list is the product that lost the most, not the one with the most units.
Money leaks is the third list: products that look fine on revenue and are quietly losing on profit, usually because returns or discount are eating the margin. This is the list a store admin cannot produce, because it needs the deductions joined to the sale.
True profit means after every deduction
The engine computes profit per item by taking the line's net sales after the discount actually realised on it, subtracting the returns that came back against it with their handling, and subtracting the cost of the unit on the sale date. Where cost is missing, the line says No cost data and is excluded from the ranking rather than ranked at a false margin.
This is why the product analyst and the net-sales bridge always agree: they are the same deductions applied at different grains. A product's net in the analyst sums to the bridge's net for the same range.
What is a brand-level verdict?
Lines roll up to brands, and each brand gets a verdict with the arithmetic shown: what it sold, what it kept, what came back, what it cost, what was left, and the coverage of cost data behind the figure. A buyer negotiating with a label sees whether that label earns its space after returns, which is a different conversation from sell-through alone.
The verdict is a computed statement, not an opinion. The AI writes the commentary around it, in the page and in the chat analyst, but the verdict's figures are the backend's and can be recomputed with Explain this.
What does the landed-cost engine add underneath?
Supplier price is not what a unit cost on the shelf. Freight, duty, clearing, handling and shrinkage sit on top of it. The landed-cost engine lets a brand record a shipment from its invoices plus a CSV of the lines, and from that sale date onward every margin figure in the product analyst uses the real landed cost, cent-exact.
Costs are dated. A shipment recorded in March does not change February's margin; it changes the cost of units sold from the date the shipment landed. That dating is what keeps past periods still, so a report the board read in April does not silently change in May.
What a buyer does with it
The winners list is the reorder candidate list, cross-checked against the inventory analyst's cover. The losers list is the markdown and drop candidate list, with returns analytics showing whether a page fix would rescue a line before it is marked down. The money leaks list is where a buyer finds the product that should never have been discounted.
Every action a buyer takes from here is proposed through Review & Send: a reorder, a markdown proposal, a note to the merchandiser. The analyst ranks and explains; the buyer signs.
Which roles read this surface on ftes.ai?
The Controller FTE owns true margin and loads the costs the store platform ignores; the Landed Cost & Compliance FTE owns the shipment records the landed-cost engine dates; the Buyer FTE reads winners and cover to propose the buy; the Pricing Officer FTE reads money leaks to find the discounts that gave margin away.
A brand that wants the product analyst working on its own range starts by connecting the store and loading costs. The first winners and losers lists will carry coverage notes until the costs are in; recording the last few shipments is the fastest way to make them complete.