How a Jewelry Brand Found a Hidden $2,429/Month Profit Leak by Tracing Loss to Two SKUs

01The challenge

A jewelry brand was down $6,340 year to date with no clear cause. Account-level P&L looked fine on the surface, organic sales made up 42-64% of the mix, and conversion was up 49% year to date, so the loss did not track with an obvious demand or listing problem.

Account-level numbers hide SKU-level problems. Without breaking the P&L down by product, the team could not tell whether the loss was systemic or concentrated in a handful of SKUs.

02The approach: The SKU-Level Loss Trace

The SKU-Level Loss Trace rebuilds profitability at the product level instead of the account level, tests the obvious hypotheses with data, and isolates the real source before touching spend.

1. Rebuild the P&L at SKU level

   Rebuilt the brand's year-to-date P&L at SKU level across Sellerboard, Advertised Product, Search Term and Business Report files (January through July 2026), isolating profit contribution per SKU rather than per account.

2. Isolate the loss-making SKUs

   Identified two hero SKUs absorbing 71% of ad spend at 170-237% Real ACOS as the sole source of the loss, and recomputed June without them to size the leak.

3. Retire the wrong hypothesis

   Data-tested and retired the branded-cannibalization theory, since branded spend was only 0.1-0.7% of total all year, and redirected the fix to spend reallocation on the two SKUs instead.

03The results

Recomputing June 2026 without the two loss-making SKUs flips the month from -$787 to +$1,642 pro forma, a $2,429 swing, recoverable through spend reallocation alone, with no new budget, creative, or catalog work required.

Why it worked: retiring the cannibalization hypothesis with data stopped the team from spending cycles on the wrong problem. Isolating the loss to two specific SKUs turned a vague year-to-date problem into a fixable, SKU-level plan.

04FAQ