Motor Aşin: a data warehouse joining two ERPs, with dashboards and decision support

2 ERPs

Sources

Old and new systems in one warehouse

Uninterrupted

Data flow

Daily, since October 2024

8+

Dashboards

From sales to purchasing

3

Phase

Setup → dashboards → decision systems

Problem

Motor Aşin’s ERP had reports, but not enough for decision support. Data lived in two systems: history in Tiger, Logo’s previous version, and current data in J-Platform. Every team pulled its own report from the ERP or from Excel with its own assumptions; the same question could have two different answers. A reporting system that would track sales, planning, supply, logistics and CRM processes in one language was needed, and a data architecture to feed it.

The introductory meeting in November 2023 went further than that: which customer from which plant, dynamic price proposals, delivery dates, purchasing optimization, network design. All of it first required a single reliable data source. The engagement was defined under four headings: data infrastructure, metrics and KPIs, Power BI reporting, training.

Phase 1: Setup and data warehouse (2024)

We started with the infrastructure: virtual servers, permissions, SSIS installation, VPN. Then the data from both sources was analysed and validated together with Motor Aşin’s team. Raw data pulled from the ERP and assumptions made in the warehouse can diverge. The teams using the dashboards know both the data and the process well, and a discrepancy they spot undermines their trust in the warehouse. So we did the validation with the process owners.

The data warehouse was built on SQL Server: product (base and attributes), dealer, customer, time, warehouse and supplier dimensions; sales, price, cost, live stock, stock movements, cancellations and returns, and order facts; budget. SSIS packages were built dynamically for the new technology; Tiger data was loaded once but packaged so it can be reloaded if needed, and J-Platform data flows intraday.

The first two dashboards came out that year: Sales, for monthly, weekly and daily analysis of four years of sales; and Live Stock, for stock on hand across the warehouses.

Phase 2: Dashboards and mentoring (2025)

So that Motor Aşin’s team could sustain the warehouse independently of Lumtify, they were trained on the SSIS setup and on data visualisation; then mentoring began. The team’s most-used reports were listed and selected, and the team built its own dashboards: Customer Region Analysis, Warehouse Fill Rate, Stock. Mid-year it became clear that mentoring alone would not hit the targets on time, so the plan was updated and development and mentoring ran together. Lumtify added the Product, Budget, CRM Salesperson Scorecard, Sales–Orders and Purchasing Performance dashboards.

A “bringing a report into the warehouse” process was defined: a meeting with the developer, a review of the existing report, a shared area for external sources such as Excel, go-live in a separate workspace, conversion to the warehouse structure, old-versus-new data validation, retirement of the old report. Reports teams had kept in spreadsheets are being moved into the warehouse one by one this way.

The second phase closed with a presentation to top management at the end of 2025; data has been flowing without error for a year.

Phase 3: Decision support (2026)

The third phase was planned along two axes. Operational excellence and stock: stock-out and lost-sales analysis, idle stock, product–supplier performance, demand forecasting, inter-warehouse transfers, product life cycle. Commercial growth and customer value: cross-sell and basket analysis, product recommendation, RFM segmentation, retention, a ranking algorithm. Alongside these, cancellation–return, productivity and CEO dashboards and a sales cube.

The plan was ranked with the team by impact, status and priority: basket analysis and stock-outs–lost sales were taken into the first programme; idle stock, supplier performance, CLTV and transfer effectiveness follow. Two senior roles were defined on Lumtify’s side: one for data flows and algorithms, the other for Power BI and mentoring.

Why this way

We chose this order on purpose: setup and a validated warehouse first, dashboards next, decision systems last. Had we started writing algorithms on day one, inconsistent data would have produced inconsistent proposals. At every step the team took on whatever it could do itself. The point of the mentoring was for the system to keep running without Lumtify.

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
• Kalekim: from MVP to a forecast management system in three years

All our work