Service · Data and analytics

Data warehouse

Think of a machine-made carpet producer in Gaziantep exporting to more than thirty countries. Orders and invoices sit in Netsis, export files and freight bills in Excel, Amazon and Etsy sales in separate seller panels, ad spend in Google Ads. Before each board meeting the finance manager loses two days stitching sheets together, and still the simplest question has no clear answer: in which country and product group do we actually make money once freight and customs are deducted? A data warehouse makes that question answerable on any given day. It changes none of your existing programs. It reads from all of them, joins the data using shared definitions and preserves history. Dashboards draw from this single source, so finance and sales stop arriving at meetings with different spreadsheets. Apply builds it remotely and then either hands the running system to your team or keeps operating it for you.

Multi-currency
lira, euro, dollar and units side by side
History kept
even when regions or prices change
Row-level access
each person sees their own patch
EU hosting
on an account in your company's name

What the work covers in practice

A warehouse earns its keep by answering the first real question quickly and beyond argument. That is why we design the scope backwards from the questions.

Agree the scope with the engineer who will do the work

Exchange rate and inflation layer

A daily rate table lives inside the warehouse. The same sale can be read in lira, euros, dollars and units, so you see how much of the lira growth reflects real volume.

Allocating costs to sales

Freight, customs, insurance, marketplace commission and advertising are assigned to the relevant order, country or product group. What gets reported is contribution margin rather than gross turnover.

Source connections

Logo, Mikro, Netsis, Nebim V3, Business Central or Odoo; ikas, Ticimax, WooCommerce, Shopify; reports from Trendyol, Hepsiburada, Amazon and Etsy; Google Ads and Meta; spreadsheets in SharePoint and Google Drive.

History stored correctly

When a dealer moves to another regional manager, past sales remain with the old region; when a price list changes, earlier periods are still read at the old price. Year-on-year comparisons stop breaking by themselves.

Automated data tests

Transformations written in dbt are tested every night for missing days, negative stock and unusual jumps. When something is off, the owner gets an alert and dashboards show the date of the last sound data.

Access and privacy

Row-level security in Power BI or Metabase: a regional manager sees their region, a dealer sees their own orders. Personal data the analysis does not need is either left out entirely or pseudonymised.

Hosting

PostgreSQL on Hetzner, OVHcloud or IONOS, or a managed database in an EU region of Azure, AWS or Google Cloud. The KVKK question of cross-border transfer is settled before anything is built.

How we approach the job, from first call to handover

Rather than a grand project plan, each step depends on the previous one being genuinely used.

01

Decision list

In one video meeting with management we list the decisions that are taken by looking at figures. Version one of the warehouse serves only those decisions.

02

Draft data model

Which field comes from which table, how each metric is calculated and which currency rules apply, gathered in one document for accounting and sales to approve.

03

First dashboard and reconciliation

The first dashboard is built from a single source and several past months are checked against the books. No new source is added until every difference is explained.

04

Usage review

We look at which dashboards people actually open. Unused ones are retired, popular ones are enriched with further sources.

Under high inflation, a lira-based report breaks a record every month. Turnover looks fifty per cent up on the same month last year while the square metres sold may have fallen. Managers who plan stock and investment from that table will get it wrong. So every amount in the warehouse is converted at the rate on the transaction date and, where possible, into units as well, and dashboards show volume by default.

Frequently asked questions

No. If your team likes Excel, it can connect straight to the warehouse. The difference is that everyone pulls from the same live table instead of stitching together private copies.

Yes. With row-level permissions each dealer sees only their orders, account balance and progress towards target. Sharing is planned around your Power BI licences or Metabase's built-in sharing.

By default the Central Bank of the Republic of Turkey selling rate on the transaction date, but if your accounts use a different source we follow that. The rule is written down and stated on the dashboard.

A data centre in the EU counts as a cross-border transfer under KVKK and brings extra steps. Keeping personal data out of the warehouse, or pseudonymising it, removes much of that burden; the legal assessment is something you make with your data protection adviser.

Build work is billed at €55 an hour plus VAT against an estimate given in advance, and a fixed-price quote is possible when the scope is clear. Hosting and Power BI licences are separate items billed directly to your company's account.

Agree on one version of the numbers

Tell us which questions go unanswered in your meetings and which programs the figures come from. The first dashboard is built around those questions.

Availability
Weekdays 09:00-18:00 Turkey time (GMT+3); an answer follows by the next working day
Calls
By video, over Microsoft Teams or Google Meet

The only cookies here are the essential ones: they keep the site running and remember the city you picked. Nothing is used for advertising or tracking. See our privacy notice for more.