Derimod business intelligence: 20+ Power BI dashboards fed from one warehouse

20+

Dashboards

One warehouse, one definition

4

Catalogue steps

Reporting → deep analysis → trading → CEO

1

Data sources

No report has its own

Daily

Refresh

Capacity and stock flow live

Problem

Had the data warehouse stayed somewhere only IT and the analytics team could reach, little would have changed for the rest of the company. At Derimod, weekly meetings began by arguing about whether the numbers were right; every team came from its own Excel and two teams’ “sales” did not agree.

The goal was for everyone who needed the data to get to it and read it directly. Whichever dashboard a figure appeared on, it had to come from the same table.

A four-step catalogue

We built the dashboards in catalogue order. Each step stands on the one before; the second does not start until the first is done.

Step	Dashboards	What it shows
1 · Reporting	Sales · Product · Customer · Sales by city	Receipts, units, revenue, cost, average price by year, week, day, product, supplier, channel and store; LFL, MTD, YTD and Pareto by product; customer count, new/repeat, frequency, cohorts
2 · Deep analysis	Basket analysis · Budget/target · Capacity	Confidence, support, lift and the product network; revenue and unit budget, AOV, IPT; last 7 days, overall and shelf fill, store + in-transit stock
3 · Trading	Trading dashboard	Cross views: regional manager × store, supplier × product group, product team × product; tracking by season group
4 · CEO	CEO dashboard	Store and product-group analysis at a glance; a summary of every layer below

Every dashboard’s metric list, pages and audience were written down before it was built; the team knew what it would see when it opened.

Operations dashboards

Alongside the catalogue are dashboards born of operational need:

• Store capacity: total capacity, actual and final stock, option capacity, number of unbroken options, stock in transit, dispatch orders, incoming stock, cover and 28-day cover; separate for apparel and footwear. Stock and capacity data flow live.
• Warehouse capacity: actual, final, available and returns stock; how much stock sits in the warehouse by season and subgroup.
• Goods-in KPIs, dispatched vs waiting in the warehouse, logistics: warehouse in/out, cost and stock pages.
• E-commerce operations, best–worst sellers, supplier performance.
• Optail reports: transfer effectiveness, replenishment success, weekly size-run breaks, idle stock.

We kept the architecture simple on purpose: all dashboards are fed from the same data warehouse. Because a sales figure comes from the same table whichever dashboard shows it, the “your report doesn’t match mine” conversation never starts. We designed each dashboard with the team that would use it. Store operations, for instance, knew best what a store capacity report should look like.

The third year

How well a BI project was built shows a few years after launch. When refresh times grew, we found and fixed the slow query instead of raising the timeout, and moved large tables to incremental refresh. Source schema changes are absorbed by the warehouse layer; the dashboards never need to know the source system’s details. Every dashboard has a business owner, which keeps the dashboard count from growing unchecked.

The change showed most in the weekly meetings. Nobody argues about the numbers any more; the meeting opens with “what do we do”. There was an effect we did not expect, too: once data was easy to reach, people asked more questions, and part of the catalogue came out of those questions. The set presented to the board in May 2024 (sales, budget, store and warehouse capacity) has been the core of management briefings since.

Related

• Derimod: seven projects in three layers, from data warehouse to Optail
• Defacto DSO: setting up a data science office and four years of consulting
• Defacto Norm Staffing: calculating the staff each store should have, from data

All our work