BI and AI Retail Transformation
Six connected workflows take a fashion retailer from last season's results to next season's buy, then to the shelf, the books and back again. One data platform, one set of numbers. Automation and Optimization.
One season, six workflows
Most retailers run these as separate departments with separate spreadsheets, so every hand-off means re-keying. On the Cybex Retail Data Platform each workflow starts from the previous one's output. Last season's results set this season's buy, and this season's results feed the next review.
- 1AI Assortment PlanningWhat worked last season, and what should we buy next?
- 2AI Merchandising & ReplenishmentPlan, buy, publish, receive, reorder.
- 3AI WMS AllocationThe right units in the right store and size.
- 4Markdown OptimizationMark down on purpose, not by the calendar.
- 5AI Sales Audit & AccountingEvery sale, tender and invoice reconciled.
- 6BI & AI Operations AnalyticsRun the business on one set of numbers.
Automated runs without anyone touching it AI the system recommends, a person approves Decision a person decides with the data in front of them BI reporting and monitoring
Score last season, then plan the next
| # | Step | Cybex | Mode |
|---|---|---|---|
| 1 | Load and reconcileSales, receipts, stock, orders and returns to vendor loaded nightly. The stock ledger is balanced to on hand before anything is analysed. | Data Platform | Automated |
| 2 | Season scorecardThis season against last: sales, gross margin, markdown rate and full-price sell-through by season code, department, class, brand and style-colour. | AI Insights: Season Review | AI |
| 3 | Breadth and depthOptions per class, units per option, and the tail of style-colours that sold one unit or none. | AI Insights | AI |
| 4 | Older stock and margin leaksWhat carried over from earlier seasons, what it sold at, and which departments lost margin. | AI Insights | AI |
| 5 | White spaceClass by price band cells where sell-through and productivity say add options, and cells that carry more options than sales. | AI Insights | AI |
| 6 | Brands to back, brands to cutSell-through, sales per option and what is already on order for each brand. | AI Insights | AI |
| 7 | Size curvesReceipts and sales by size family, and sizes bought one unit deep. | AI Insights | AI |
| 8 | The buy gapNext season's orders against the same date last year, the units needed at a target sell-through, and how much buy is left. | AI Insights | AI |
| 9 | Assortment briefA written summary for the buy meeting: where to add, where to cut, which brands, which sizes, how much is left to buy. | Executive report | Decision |
Efficiency
A full season review in hours rather than weeks of spreadsheets, built the same way every season.
Optimization
Fewer, deeper options. The buy is sized to a sell-through target instead of last year plus a percentage.
Automation
The dataset recalculates every night. The live view and the executive report come from the same numbers.
Plan, buy, publish, receive, reorder
| # | Step | Cybex | Mode |
|---|---|---|---|
| 1 | New season merchandise planSales, stock-to-sales and markdown % by class and month, started from last year and the season review instead of a blank sheet. | Merchandise Plan: New Plan | AI |
| 2 | Open-to-buyOn order is refreshed every night from the purchase orders, so OTB by class and month is always current. | Merchandise Plan: OTB | Automated |
| 3 | Order workbenchLast season's winners to rebuy, new styles matched to the style they replace, size curves applied, draft POs built inside OTB. | Merch Plan Style: Order Workbench | AI |
| 4 | Create new productStyles, colours and size SKUs generated from the size table, with attributes, prices and prints on blanks. | Product maintenance | Automated |
| 5 | Approve and issueDraft POs reviewed and released to vendors as purchase orders. | Purchasing | Decision |
| 6 | Publish to e-commerceProduct, variants, prices and status published to Shopify from the same product master, so nothing is keyed twice. | Shopify publish | Automated |
| 7 | ReceiveReceipt scanned against the PO, price tickets and GS1-128 labels printed, over, short and substitutions flagged on the spot. | Receiving and ticketing | Automated |
| 8 | In-season opportunitiesReorders and chases on winners from the forecast (six-week base and weeks of cover), plus direct-to-store buys. | Suggest PO, Direct-to-Store | AI |
Efficiency
Plan, OTB, POs, product and the online store come from one product master. Product is set up once.
Optimization
Every PO is checked against open-to-buy and the season review. Winners get chased while there is still season left.
Automation
Nightly on-order, generated SKUs, one-click publishing to Shopify, labels at receiving.
The right units in the right store and size
| # | Step | Cybex | Mode |
|---|---|---|---|
| 1 | Initial allocationEach DC receipt split by store from the store's forecast need, 1 to 8 weeks of cover, by size. | AI Allocation | AI |
| 2 | Pick, pack and labelPick lists, box counts and carton labels generated from the allocation. | WMS, shipping | Automated |
| 3 | Ship and receive at storeTransfers sent and confirmed at the store, so stock in transit is never lost between systems. | WMS, POS | Automated |
| 4 | Replenish from the DCUnits sold are replaced from DC stock each week, up to each store's cover target. | Replenishment | Automated |
| 5 | Re-allocate between storesWhen the DC is out, slow stores give to stores that sell, by size. | Re-allocation | AI |
| 6 | Consolidate before markdownBroken size runs gathered into fewer stores, so they sell before they need a markdown. | Re-allocation | Decision |
| 7 | DC productivityUnits in and out, flow by day, open allocations. | DC analytics | BI |
Efficiency
Allocators review exceptions instead of building every allocation by hand.
Optimization
Stock follows demand by store and size, which means fewer lost sales in some stores and fewer markdowns in others.
Automation
Allocation, pick lists, labels and weekly replenishment run from the same forecast.
Mark down on purpose, not by the calendar
| # | Step | Cybex | Mode |
|---|---|---|---|
| 1 | Where each style standsSell-through against plan, weeks of cover, and weeks since first receipt, by style-colour. | AI Insights | AI |
| 2 | CandidatesSlow sellers, broken size runs and older season codes, ranked by retail value at risk. | AI Insights | AI |
| 3 | Timing and depthAn early, shallow markdown against a late, deep one, compared by class on margin dollars. | Markdown model | Decision |
| 4 | Price changeOne approved change sent to POS and e-commerce together. | Price change, Shopify | Automated |
| 5 | MeasureLift, margin dollars and markdown rate against plan, week by week. | BI | BI |
| 6 | Feed the planActual markdown % returns to the merchandise plan and next season's review. | Merchandise Plan | Automated |
Efficiency
One weekly list instead of a store walk and a spreadsheet.
Optimization
Markdowns go where they move units, so more of the season sells at full price.
Automation
Approved prices reach the stores and the website together, and results are measured on their own.
Every sale, tender and invoice reconciled
| # | Step | Cybex | Mode |
|---|---|---|---|
| 1 | Daily closeStore and web orders loaded and balanced: tickets, tenders, taxes and returns. | Sales Audit | Automated |
| 2 | Card matchProcessor settlement matched to POS payments by store, register and invoice. | Card reconciliation | Automated |
| 3 | Exceptions explainedTerminal logs read for every unmatched payment: lost terminal reply, rung twice, approved but not settled, partial approval. | AI Sales Audit | AI |
| 4 | DepositsCash and card deposits balanced to the bank. | Deposit balancing | Automated |
| 5 | Receipt, PO and invoice matchVendor invoice checked against PO cost and received quantity, within tolerances. Only the differences go to AP. | Three-way match | Automated |
| 6 | Inventory at costStock ledger at moving average cost: receipts, transfers, adjustments, shrink and markdowns for any period. | Stock ledger | Automated |
| 7 | GL journalSales, tenders, cost of sales, inventory and payables posted to the ERP, such as Business Central. | GL journal | Automated |
Efficiency
Auditors and AP clerks work the exception list, not every transaction.
Optimization
Lost card payments, duplicate charges and vendor overbilling are caught within days, not at year end.
Automation
Daily matching, deposits, the stock ledger and the GL journal run every night.
Run the business on one set of numbers
| # | Step | Cybex | Mode |
|---|---|---|---|
| 1 | Store operationsTraffic, conversion, average ticket, units per transaction and sales by associate, daily and weekly. | BI templates | BI |
| 2 | Merchandise analyticsSell-through, cover, aged stock, and size and brand matrices for any period. | BI Query and Matrix | BI |
| 3 | AI InsightsFindings and actions in plain language, generated from governed datasets, with the data behind every number. | AI Insights | AI |
| 4 | Exceptions to ownersAudit, stock and plan variances sent to the person who acts on them. | Alerts | Automated |
| 5 | AI data warehouseEach insight's dataset stored by season, building the history that later models learn from. | AI data warehouse | Automated |
| 6 | Close the loopThe season's results become step 1 of the next season review. | All six workflows | Decision |
Efficiency
No month-end report builds. Every role opens the same numbers each morning.
Optimization
Store, merchandise and finance results side by side, so the cause of a problem shows up, not just the symptom.
Automation
Datasets, insights and alerts refresh overnight, and nobody has to export anything.
Back-office automation: from purchase order to general ledger
The biggest efficiency gain is the paperwork between the workflows. One product master and one transaction record carry each unit from the order to the books, with no re-keying.
- Purchase orderBuilt in the workbench, inside OTBWorkflow 2
- ReceiptScanned against the PO, over and short flaggedWorkflow 2
- Merchandise handlingTickets, GS1-128 labels, cartonsWorkflows 2 and 3
- Allocation and transferDC to store, store to storeWorkflow 3
- Invoice matchInvoice vs PO cost vs received quantityWorkflow 5
- Inventory at costMoving average stock ledgerWorkflow 5
- GL journalPosted to the ERP, balancedWorkflow 5
Technology adoption: five stages, each paying for the next
Retailers don't need to switch everything on at once. Each stage of the Cybex Retail Stack runs on the data you already have, and each one delivers results before the next begins.
One reconciled dataset
Connect POS, e-commerce, ERP and WMS to the Cybex Retail Data Platform on your existing data warehouse. Balance sales and stock to the books.
One set of numbers
BI templates for stores, merchandise and audit replace exported spreadsheets. Everyone reads the same figures every morning.
Insights with actions
Start with a season review for the next buy. It shows value inside one buying cycle.
Remove the re-keying
OTB, product publishing, receiving, card matching, invoice matching and the GL journal run on their own.
Optimize to targets
The buy, allocation and markdowns are set against sell-through, cover and margin targets, and measured every season.
How each role changes
| Role | Today | With the six workflows |
|---|---|---|
| Planner | Builds the plan in spreadsheets from exported reports | Starts from last year and the season review. OTB is current every morning |
| Buyer | Prepares line reviews by hand | Works from winners, white space, brand scores and draft POs in the workbench |
| Product and e-commerce | Keys product into the POS and again into the online store | Creates product once and publishes it to Shopify from the same master |
| Allocator and DC | Splits receipts by hand and reacts to store calls | Reviews AI allocations and re-allocations; labels and pick lists are generated |
| Sales audit and AP | Ticks and ties every tender and invoice | Works only the exception list, each exception with its likely cause |
| Finance | Waits for month end, then reconciles | Receives a balanced daily journal and inventory at cost for any period |
| Executive | Reads a monthly deck | Reads nightly insights with the actions that go with them |