Executive answers from governed retail data.
Two kinds of insight over one data platform. Static pages refresh with the nightly load. Dynamic pages are generated on demand from a dataset and a written brief, so a new question on Monday has a challengeable, repeatable answer the same day.
What the analysis found, and how it is produced.
Each item is a finished piece of work on retail data, not a demo. Case studies and dashboards are static; the engine page shows how the dynamic pages are made.
The dynamic insights engine
A governed dataset plus a written brief produces an executive page: white space, size balance, DC cost per shipped unit, cashier exceptions.
How it works → 02 · CASE STUDYWhy growth slowed in week 27
Growth halved in one week while traffic rose. Conversion fell, and the cause was availability: stock in the wrong stores.
Read the case study → 03 · DASHBOARDWhitespace analysis
Sales, margin and growth concentration for a clothing division, with the reallocation signals an executive can act on.
Open the dashboard → 04 · CHECKLISTTen reasons to automate sales audit
Efficiency, accuracy, loss prevention and faster decisions when the audit runs on the data instead of on paper.
Read the checklist → 05 · PROOFProof in production
What runs today at a Canadian specialty retail group, on the retailer’s own platform, and what was added in the last twelve months.
What is live → 06 · THE CYBEX QUARTERLYTwenty-four essays
Retail AI strategy, allocation, merchandising, sales audit, customer intelligence, operations, pricing, margin and platform deployment.
Read the Quarterly → 07 · EBOOKMerchandise Planning & Inventory Optimization
An integrated operating framework from assortment intent to executable stock policy.
Open the ebook → 08 · BI ANALYTICS SERIESThe daily routine
The governed report packs the insights read: matrix, list, query and dashboard templates with a guide behind every Note button.
The series →A brief for every retail function.
Sixteen briefs are written today. Each names the audience, the grain of the dataset, what every column means, the formulas, the thresholds that define a finding and the order of the page.
White space and over-space
Class by region against the plan: growth, markdown, stock share versus sales share, open orders against the run rate.
Merchandising → ASSORTMENT & SIZESize curve balance
Sell-through by size within style-colour, the spread that flags an imbalance, and open-order corrections by size in units.
Assortment → DISTRIBUTION & DCDC productivity and store lanes
Touch units per hour, cost per shipped unit, idle days, stock draw-down; units per document and gaps between shipments by store.
Allocation & DC → SALES AUDIT & LPWeekly exceptions and margin
Cashier watch list, after-hours returns, drawer opens, over/short by till; the margin bridge from full retail to net sales.
Loss prevention → STORE OPERATIONSTraffic, conversion, staffing
Four actual weeks and four forecast weeks by store: visits, conversion, sales per staff hour and the hours the forecast implies.
Store analytics → PRIVATE LABELBlank requirement and production schedule
Private label demand rolled to the blank and colour over 26 and 52 weeks, the production schedule limited to blanks on hand, the reorder.
Private label →Dataset, brief, page. Then challenge it and run it again.
The definitions live in the brief, not in the model’s head. An executive can dispute a threshold, an analyst changes one line, and next week’s page is comparable to this week’s.
Governed dataset
One documented view per function, sized so the whole result fits a single call.
Write the definitions
Grain, horizon, column meanings, flows versus positions, formulas, thresholds, page order.
Produce the page
Executive summary, scorecard, tables, ranked findings with an action each, data notes.
Argue with the evidence
Every finding carries its numbers in one line, so the ranking can be disputed.
Same brief, next week
A coached cashier, a rebalanced size curve or a moved buy is visible against last week.
Nothing acts on its own
Pages recommend; people approve; the applications act on what is approved.
Bring one question. Leave with a page.
A working session on your data ends with a brief written and a first page generated. A Data Platform Foundation pilot ends with two BI series and one executive insight in production.