Case studies

Evidence, filterable.

6 completed engagements, reported the way we report to clients — stat first, method shown, constraints included. No logos; the category runs on NDAs, and the numbers don't need them.

System

Marketplace

Revenue band

Showing 6 of 6 cases · constraints included in all 6

OPS / FULFILLMENT / 04DemandProfitUS8-fig+31% contribution margin in 2 quartersReactive reordering replaced; ad spend re-weighted to margin.Constraint: hero SKU capacity-limited through Q3Read the system OPS / PPC DESK / 07AdvertisingUS7-figTACOS 14.1% → 8.6% in 2 quartersBid ceilings rebuilt from margin; revenue flat 1 quarter while low-margin spend unwound.Constraint: ramp took the full first quarterRead the system OPS / PORTFOLIO / 11ReportingEU-58-fig1 reporting layer across 23 P&LsEBITDA variance between brands cut from 19 pts to 7.Constraint: 2 brands exited the programRead the system OPS / CONTENT DESK / 09ListingUK7-fig+24% CVR on the hero SKU in 1 quarter7 tests run, ranked by impact × confidence.Constraint: 3 of 7 tests lostRead the system OPS / PRICING DESK / 02ProfitUS8-fig+4.1 pts contribution margin, TTMA third of the bridge came from fee recovery nobody was watching.Constraint: 1 price step reversed — loggedRead the system OPS / WAREHOUSE / 12DemandUS7-fig0 stockout days across Q4 peakPeak forecast built from 2 years of velocity + supplier variance.Constraint: 12% overstock accepted as insuranceRead the system

No cases match that combination yet.

We publish every completed engagement — this one hasn't happened. When it does, the constraints will be in it.

Case 04 — FBA · US · 8-fig home brand · systems installed: Profit & Pricing → Demand & Inventory

+31% contribution margin in 2 quarters.

The situation — as found, onboarding audit

0Stockout days per quarter

0%Of COGS — air freight rescues

0Quarters of flat profit on +20% revenue

A $14M home-goods brand growing 20% a year and keeping none of it. Reorders were chased after the stockout; air freight rescued revenue and donated the margin; ads were optimized to an ACOS target that treated every SKU as the same customer.

The founders had fired 2 agencies in 18 months. The brief they gave us was 1 sentence: "Tell us where the profit goes."

The system — installed over 90 days

2 systems, 1 drawn interlock.

01Profit & Pricing — weeks 1–4Fee audit + unit economics current. The margin floor first, so every later decision had a true denominator.

02Demand & Inventory — weeks 3–10Daily forecast per SKU, reorder points with supplier variance, human approval gate, automated POs.

03Ad re-weighting — weeks 8–12Bids re-anchored to margin per SKU, stock-gated by the demand system. No new ad platform — new arithmetic.

The result — 2 quarters, reported monthly

Reported with the misses on the chart.

Contribution margin+0%Q1 → Q3 · settlement data

Stockout days/quarter31 → 0Q3 vs onboarding

Air freight share of COGS11% → 0%Q3 vs onboarding

Forecast accuracy0%Q2 · published · launches excluded

Forecast vs actual — contribution margin by month

ActualForecast — dashed, always

The misses are on the chart. Months 2 and 4 came in under forecast — they're why the Q2 accuracy number means something.

Source: client scorecard · SP-API settlement data · Jan–Jun 2026

Constraints — what didn't work

The hero SKU stayed capacity-limited through Q3 — supplier MOQ, not forecast. Margin gains came from the other 41 SKUs.
1 price step was reversed after 11 days — elasticity said no. The reversal is in the log, and in this report.
New launches ran manual all 6 months. No history, no autopilot — the 2 launched SKUs are excluded from the accuracy number.

"First partner that told us what wouldn't work — then proved what would."

COO — the operator in this case

Method — numbers from client scorecards and SP-API settlement data · periods as labeled · exclusions stated above

The next case study could be your account.

The audit is where every one of these started — 14 days, read-only, and the report is yours to keep.

Book a systems audit