Commerce & Retail
Brands and retailers selling through their own storefronts, marketplaces and stores, with stock, returns and promotions to control.
01 · The operating chain
The commerce & retail operating chain
Each demonstration sits at a point in this chain where a real business loses information, margin or time. The chain is how we read a business; the demonstrations are how we prove we can act on it.
01
Acquire
1 demonstration
02
Merchandise
1 demonstration
03
Sell
4 demonstrations
04
Stock
2 demonstrations
05
Fulfil
1 demonstration
06
Return
2 demonstrations
07
Reconcile
3 demonstrations
08
Retain
1 demonstration
Coverage map · demonstrations by operating area
| Area | Working demonstrations | Count |
|---|---|---|
| Acquire | Growth Quality Investigator | 1 |
| Merchandise | Promotion Profit & Inventory Guard | 1 |
| Sell | Commerce Leakage Investigator · Inventory Truth & Oversell Prevention · Promotion Profit & Inventory Guard · Customer Promise Integrity System | 4 |
| Stock | Inventory Truth & Oversell Prevention · Return Disposition & Value Recovery | 2 |
| Fulfil | Customer Promise Integrity System | 1 |
| Return | Commerce Leakage Investigator · Return Disposition & Value Recovery | 2 |
| Reconcile | Commerce Leakage Investigator · Inventory Truth & Oversell Prevention · Settlement & Revenue Assurance Investigator | 3 |
| Retain | Growth Quality Investigator | 1 |
A demonstration shows understanding of an operating area on a fictional business. It is not a record of client work.
02 · Where it breaks down
Where commerce & retail breaks down
- 01Healthy revenue hiding falling contribution
- 02Several systems disagreeing about what is in stock
- 03Promotions run on instinct rather than unit economics
03 · Working demonstrations · 7
Working demonstrations
contribution leakage
Commerce Leakage Investigator
A synthetic direct-to-consumer brand whose sales look healthy while returns rise and contribution slips. The system reconstructs the numbers, traces returns to the supplier batch they shipped from, compares specifications, retrieves what customers said, proposes corrective actions under the business's own approval policy and executes the approved ones against the session's records.
Runs on a generated dataset (Halden & Co., a fictional brand). Every customer, order, supplier and figure is synthetic and labelled as such in the interface.
inventory reconciliation
Inventory Truth & Oversell Prevention
A synthetic retailer's storefront, marketplace, ERP and warehouse each report a different available-to-sell figure. The system reconciles every SKU against the physical truth, decomposes each gap into the causes the records support (returns shown as available before inspection; picks lost in a failed export window; a pending cycle count), lists the orders already accepted against stock that is not there and the promotions still driving demand, and executes corrective actions under the retailer's own policy.
Runs on a generated retailer (Marlowe & Finch, a fictional brand). Every system record, order, return and integration event is synthetic and labelled as such in the interface.
promotion economics
Promotion Profit & Inventory Guard
A synthetic merchant has drafted a 25%-off, free-delivery campaign across twenty lines. Before it runs, the system computes every line's unit contribution at full price and under the promotion from its own cost of goods, fees, fulfilment, delivery class and return history, expects units from the line's recent sales and the category's measured promotion response, checks stock and inbound cover, and gives each line a decision with its reason. Three plans are derived from the same arithmetic; the commercial lead approves one and the campaign configuration genuinely changes.
Runs on a generated merchant (Sable & Oak, a fictional brand). Every product, sale, return, cost and stock figure is synthetic and labelled as such in the interface.
acquisition economics
Growth Quality Investigator
A synthetic homewares merchant acquired one quarter of new customers through six sources and observed every cohort for five months. The system reconstructs each cohort from headline revenue to realised contribution: offer discounts, refunds and return handling, cost of goods, fulfilment, delivery subsidy, fees, support contacts and the repeat orders that actually happened. The source with the best return on ad spend turns out to be the worst buy; a high-return source is fine; the true winner sits mid-table on the report.
Runs on a generated merchant (Ferran & Vale, fictional). Every campaign, customer, order, return and support contact is synthetic and labelled as such in the interface. Nothing beyond the observed window is estimated.
delivery promise
Customer Promise Integrity System
A synthetic homewares merchant shipping from two warehouses with three carriers, four weeks of orders. Every order is re-run through the checkout rule that promised it and the calendar that had to keep it: warehouse cutoff and pick lead, carrier collection timetable, lane transit, weekends and holidays. The system separates ordinary misses from promises that were impossible at the moment they were shown, and finds them all in one combination: a generic next-day rule, one site's Friday cutoff, one carrier's earlier Friday van and no Saturday delivery. The slowest carrier has the best breach rate because it quotes a day of buffer.
Runs on a generated merchant (Orrin Home, fictional). Every order, warehouse calendar, carrier timetable, lane and delivery is synthetic and labelled as such in the interface.
value recovery
Return Disposition & Value Recovery
A synthetic outdoor retailer's returns centre receives one week of returns: sixteen products, four vendors, graded on arrival, handled by a blanket rule (grade A restock, everything else liquidate, cheap items dispose). The system prices every path for every unit from its own records: restock, refurbish, return to vendor, outlet, liquidate, dispose, using grade, reason, vendor terms and window, stock cover, season and handling costs. The rule destroys most value on opened-like-new units of short-stock lines and on defective units a vendor would credit; for overstocked last-season basics it is already right.
Runs on a generated retailer (Halden Outdoor, fictional). Every product, return, grade, vendor term and price is synthetic and labelled as such in the interface. Recovery figures are expected values at today's prices, not outcomes.
revenue assurance
Settlement & Revenue Assurance Investigator
A synthetic homeware brand selling through its own web shop and a marketplace, four weeks of orders. Every order is followed from sale to bank: captures, refunds and partial refunds, the fees each schedule allows, the provider's payouts, the marketplace's statements and settlements, chargebacks and bank credits. Every difference is classified as timing, expected fee, valid refund, unexplained difference, duplicate deduction or settlement shortfall and traced to its source records. The largest gap is timing and needs nothing; four small, specific mechanisms need a person.
Runs on a generated brand (Corvane Living, fictional) with a fictional payment provider and marketplace. Every order, capture, refund, statement line, settlement and chargeback is synthetic and labelled as such in the interface.
These are ways of working, not products.
These demonstrations show how we approach a problem; they are not a list of pre-built answers. Your processes, systems, constraints and blind spots will be different. A Business Systems Discovery starts from how your business actually operates, finds where information and decisions break down, and recommends the right mix of AI, automation, integration, intelligent systems or plain process change.