Data Department helps consumer goods companies and manufacturers in Türkiye and beyond apply machine learning (ML) and AI (artificial intelligence) to specific decisions: what to make, what to stock, which outlets need attention, what to inspect and how to schedule. We start from the decision, deliver on your own infrastructure and integrate with your ERP (enterprise resource planning) and MES (manufacturing execution system).
Where does AI pay off in consumer goods and manufacturing?
AI pays off where a decision is made often, with real money at stake, and the data already exists. In FMCG (fast-moving consumer goods), that is usually demand and the sales channel. In the plant, it is planning, stock, quality and documents.
| Decision | What ML or AI does | Typical data |
|---|---|---|
| How much to produce, and when | Demand forecasting with uncertainty ranges, by product, customer and channel | Sales orders, shipments, distributor sell-out, promotions, seasonality |
| Which outlets or assets will underperform | Early prediction so field teams can redeploy assets or target sales | Outlet sales, asset placement, visit history |
| How much raw material and spare parts to hold | Probabilistic forecasts that set stock levels by acceptable risk | Consumption history, lead times, bill of materials, stock levels |
| Whether a part or pack passes inspection | Computer vision that flags defects from camera images | Labelled images of good and defective items |
| Which machine runs which job, in what order | Optimisation (mixed-integer programming) for scheduling under capacity and due dates | Work orders, routings, machine capacity, set-up times |
| Whether a machine or process is drifting | Anomaly detection that flags unusual behaviour early | Sensor readings, process parameters, maintenance logs |
| Where an answer sits in technical documents | A RAG (retrieval-augmented generation) assistant that answers from your own documents, with sources | Procedures, specifications, quality records, supplier documents |
Relevant case studies
- Outlet cooler performance: a beverage bottler. Our model predicted which outlets' coolers would underperform, ahead of the company's own evaluation cycle, so field teams could act earlier. It reached 99% accuracy, with an estimated ~2.35% annual sales upside.
- Spare parts inventory forecasting: the probabilistic methods from our automotive distribution work, which gave ~30% better forecast accuracy and ~10% lower inventory holding cost, apply directly to raw materials and maintenance parts. See our automotive page.
What are typical first projects?
A good first project has a clear owner, existing data and a result you can measure in weeks:
- Demand forecasting for production planning, to reduce rushed changeovers and excess finished goods. See our demand forecasting service.
- Outlet or channel performance prediction for consumer brands with field sales teams.
- Raw material and spare parts stock optimisation, with stock set by risk, not habit.
- Visual quality inspection on one line: one defect type at one station, using images you can already capture. See computer vision quality inspection.
- A document assistant for technical and quality teams.
Our article on how to choose your first AI project explains how we rank candidates. The free 30-minute AI opportunity review is a quick way to test which of these fits.
Data and integration notes
- ERP holds orders, stock, bills of materials and lead times. Results are written back so planners use them where they already work.
- MES and shop-floor data sit in the MES or in SCADA (supervisory control and data acquisition) systems. We connect read-only where possible.
- Images and sensors. Cameras, sensors and installation are out of our scope; we cover models, software and integration, and work with your hardware supplier.
- Infrastructure. We deliver on AWS (Amazon Web Services), Google Cloud, Azure or on-premises servers.
Support programmes
Turkish manufacturers can apply to programmes that support digital transformation and AI, including those run by TÜBİTAK (the Scientific and Technological Research Council of Türkiye) and KOSGEB (the Small and Medium Enterprises Development Organisation), the Istanbul Chamber of Industry's İSO Dijital AI programme and AI EDIH Türkiye, a European Digital Innovation Hub. We can help scope projects for these programmes; please confirm eligibility and funding rules with each programme directly.
Frequently asked questions
Is AI realistic for a small or mid-size manufacturer?
Yes, if the project is tied to one decision and uses data you already collect. A two-week Discovery shows which use case is worth doing.
What is different about AI for consumer goods brands?
Much of the value sits outside the plant: in channel demand, promotions and outlet performance. Our beverage bottler work predicted underperforming outlet coolers before the company's own review.
Can the models run on our own servers?
Yes. We deliver on-premises or on AWS, Google Cloud or Azure, depending on your policies.
Do you supply cameras for visual inspection?
No. Hardware is out of scope. We build the models, software and integration, and work with your chosen hardware supplier.