Amica: How to Plan a Supply Chain for 20,000 Containers and 6,500 SKUs

In this episode of Rozmowy Logistyków, Piotr Skobało talks with Dawid Kujaczyński of Amica about supply chain planning at large scale. They discuss what the planning process looks like at a company with 15 warehouses in 9 countries, importing the equivalent of 20,000 containers a year from Asia to Europe.

ONE MARKET, ONE BUSINESS MODEL — AND NO TWO ALIKE

Amica manages around 6,500 SKUs, sells nearly 4 million appliance units a year, and imports the equivalent of 20,000 containers through a network of 15 warehouses across a dozen-plus countries — but the operational complexity doesn't come from the numbers alone, it comes from every market running on a different model. In the UK, the company works mainly with kitchen-studio partners: orders come in with plenty of lead time, but require practically 100% on-time delivery, because a customer can't receive half a kitchen set. In Germany, most deliveries go straight to the customer, bypassing the company's own warehouses, which makes on-time performance an especially sensitive point. On top of that come four-month import lead times from China, combining manufacturing and transport. The takeaway: scale on its own doesn't generate complexity — the variety of models that have to be served simultaneously does.

S&OP ON A MONTHLY CYCLE, PLANNING ON A 12-MONTH HORIZON

Amica's S&OP process runs in a cascade: first the local level (demand, production, and inventory plans per market), then aggregation to the global level. The planning horizon is a rolling 12 months, with heavier emphasis on the 4-5 month window, because that's the forecast against which actual orders are placed with suppliers and the company's own factory. The longer horizon (months 6 through 12) is used to prepare suppliers and the factory for demand swings and to secure transport availability — containers and space on ships. A key point: planning strategic purchasing (negotiating contracts and prices) needs a different time horizon than fulfilling a current order — companies that focus purely on the short term lose the ability to look for alternatives and better terms.

TWO FORECAST-ACCURACY METRICS, MEASURED OVER TWO HORIZONS

Amica measures forecast accuracy with two metrics: MAPE (weighted mean forecast error at the individual SKU level, where the direction of the error doesn't matter) and forecast bias (whether the organization has a systematic tendency to over- or under-forecast demand). Both metrics are measured over both a four-month and a one-month horizon — because measuring accuracy at only one point in time (e.g. exactly when the order needs to be placed) says nothing about whether the process is actually improving over time.

PORTFOLIO AND MASTER DATA MANAGEMENT AS AN UNDERRATED FOUNDATION

For about a year and a half, product lifecycle management (phasing out low-annual-turnover SKUs that generate a disproportionately large forecast error and raise the level of "unhealthy" inventory) has become a separate, important part of the S&OP process. In parallel, the company started measuring the quality of its master data (lead times, MOQs, and other planning parameters) — without which, as Kujaczyński put it, the planning department "would be practically blind." Because Amica grows mainly through acquiring local companies, every rollout of a shared system (currently a migration from SAP to SAP HANA) requires thoroughly cleaning up master data from scratch in every acquired market.

INVENTORY AND TRANSPORT PLANNING AS ONE INSEPARABLE PROCESS

Inventory levels are planned annually per market, matched to the local business model, and then verified globally against cash flow — if sales grow beyond budget, you have to check whether inventory is keeping up, and vice versa. Transport can't be planned separately from inventory: with a transport cost of hundreds of dollars per container holding around 100 refrigerators, every change in freight rate directly affects unit product cost — and in recent years, freight prices have swung by as much as a factor of ten. Seasonality adds another layer: some categories (e.g. refrigerators) sell 70% of their annual volume within two or three months, and Chinese New Year shuts down supplier factories for two weeks, requiring deliveries to be planned well ahead around that calendar.

GEOPOLITICS AND MULTI-TIER RISK (TIER 1, TIER 2, TIER 3)

The Suez Canal blockage, US-China trade tensions, and shrinking container availability (shipping lines scrapped just a dozen-plus thousand units a year in recent years, versus roughly 300,000 pre-pandemic, since the fleet now sails longer, roundabout routes) show that even companies buying locally in Europe aren't immune to global disruption — if their European suppliers themselves import components from China. Amica monitors resource availability not just at direct suppliers (tier 1), but also at their suppliers (tier 2) for key components — mirroring practices from the automotive industry.

TECHNOLOGY: FROM MANUAL COPY-PASTING TO MANAGING EXCEPTIONS

Beyond SAP and an IBP layer for integrated sales planning, Amica built its own platform connecting all suppliers and carriers in a single environment with real-time data — built almost by accident during the pandemic, when supply disruptions forced better visibility. The result: one person now manages 200-300 containers at once, focusing purely on deviations from plan, instead of manually copying data from carriers' Excel reports — without the automation, as Kujaczyński notes, the department "would spend three-quarters of its time copying data." Artificial intelligence is so far tested selectively (comparing AI estimates against sales teams' expert judgment in demand planning) — the priority remains solid process automation, not AI as a goal in itself.