Service · Data and analytics

Data warehouse

A data warehouse earns its place when the figures sleep in several systems and somebody glues them together in a spreadsheet each month. It replaces neither Sage nor Cegid nor your shop back office: it draws from them, puts everything into one shape and keeps it in one place, history included. Dashboards then sit on that single base, so sales, finance and marketing finally read the same lines.

1 database
behind every report
EU region
chosen for hosting
History
kept month after month
Every night
loading runs on its own

Where this service reaches

The data model is designed only once we know which questions the reports must answer and who will genuinely read them.

Put your scope to an engineer

Opening questions

What the report must show, who reads it and which decision it triggers. The list of questions your management asks is the real specification.

Sources

Sage, Cegid, EBP, Odoo or Pennylane, your PrestaShop, WooCommerce or Shopify store, marketplace statements, advertising platforms and the spreadsheets nobody can do without.

Platform in Europe

A PostgreSQL database with OVHcloud, Scaleway or Clever Cloud, or the managed service you already run, with access tied to your existing company accounts.

Shared model

Customer, article and sales channel linked across systems, so that a marketplace order and its matching invoice form one single transaction.

Traceable transformations

Calculation rules are written as version controlled code, with dbt for instance, and tests halt the load when a value falls outside the expected range.

Full cost of sale

Commission, advertising, carriage and returns are set against revenue, without which the margin on screen is an imaginary margin.

Reports and rights

Dashboards in Power BI or Metabase, opened by department and by role, readable from a phone while travelling.

The way an engagement runs

The date of the first report is given after we have looked at the sources. Later ones come faster, since they reuse the same foundation and the same definitions.

01

Choosing the first report

We begin with whichever report eats the most manual work today, most often margin by sales channel.

02

Design

Data model, matching of customers and articles across systems, and a schedule for the overnight loads.

03

Build and reconcile

Loading, the report itself, then a comparison with the figures finance has produced until now. Every difference is explained one at a time.

04

Further areas

Stock, receivables, marketing. Each new report attaches to the existing model instead of starting from nothing.

Revenue on a marketplace is easy to calculate; margin is far harder. The seller statement shows sales, but commission, promotion options, advertising campaigns, delivery charges and returns settle separately and often in a different month. Only once they are set against the purchase price held in your ERP can you see which listings actually earn money and which merely inflate turnover.

Questions and answers

Licence and hosting costs depend on how many people open the reports and on how much data you keep. We take the current vendor pricing and put it in the quotation before the start, separating what is fixed from what varies. Building the warehouse is quoted as a fixed price or at EUR 95/hour excl. VAT with an agreed ceiling.

Yes, and we encourage it. You receive a documented data model, and one or two people from your team take a short course in Power BI or Metabase. Loading and adapting to changes in the source systems can stay with us under a monthly contract, while your own analysis moves without waiting for us.

That is one of the points of the exercise. Reports query the warehouse model rather than the source system, so a migration means replacing one load. Earlier history stays available and year on year comparisons keep working, which is never the case when everything rests on exports.

For management reporting, once a night covers the great majority of needs. Loading more often costs more and puts pressure on source systems during working hours. Stock levels and marketplace orders are the exception and can refresh hourly when the pace of selling genuinely calls for it.

Let us build your data warehouse

Describe the reports still assembled by hand and the systems the figures come from. We will build the first report and reconcile it with your own calculations.

When we are around
Weekdays, 8:00 to 18:00 CET; answers land inside one working day
Talking it through
A call on Teams or Google Meet, whenever writing is not enough

We set strictly necessary cookies only: they keep the site running and remember the city you chose. Nothing here is used for advertising or tracking. More in our privacy policy.