PartsOS Forecast
Haver & Boecker: AI-Powered Spare Parts Forecasting for 27,000 SKUs, Service-Driven
How the global machine builder is moving from reactive, manual inventory planning to AI-based demand forecasts with PartsOS Forecast, with the goal of raising its service level.
Who is Haver & Boecker?
HAVER & BOECKER is a long-established, family-run industrial company based in Oelde (Westphalia), founded in 1887. Today it is one of the leading international technology providers in wire mesh and mechanical engineering, with subsidiaries around the world. HAVER & BOECKER brings two core divisions under one roof:
- Wire Weaving Division
- Machinery Division
Within the Machinery Division, HAVER & BOECKER develops systems for filling, packing and handling bulk materials (e.g. cement, building materials, chemicals or food). These include filling machines, weighing and dosing systems, and complete packaging lines.
Spare parts and service are a key part of this business:
- High plant availability: Spare parts are essential to avoid downtime and get production running again quickly.
- Long life cycles: Industrial plants often run for many years. A reliable spare parts supply secures long-term operation.
- Original parts & quality: Precisely fitting original spare parts ensure optimal performance, safety, and efficiency.
This makes the spare parts supply the foundation of the business for Haver & Boecker and for each of its customers worldwide. With a high share of custom machinery and long operating lives, the number of spare parts is very high. Around 27,000 different items were sold over the past five years, with around 10,000 of them sold more than five times.
The Starting Point
Planning for all items is automated according to strict, rigid rules. All items are treated the same, whether they are a spare part or a part for a new machine. There is no clear focus on spare parts and no differentiated forecasting logic: selected parts are planned with the same method, a slightly adjusted average of the past twelve months' consumption.
In practice, this meant:
- Reactive forecasts: Stock levels were based on consumption over the last twelve months.
- 27,000 SKUs, high complexity: A custom-machinery portfolio defies standardization. Stocking decisions are made across the board or adjusted by hand, which takes a lot of coordination.
- Rigid planning rules: These led to frequent small orders with process costs of around €100 each.
- Stock-outs: Critical spare parts are repeatedly missing, which extends lead times and causes delays.
- Low stock coverage and long lead times: Customers are unhappy with delivery times, which leads to lost sales, especially for high-demand items.
As a result, the planning team had to chase spare parts far too often and spent its days fighting backlogs and stock-outs. A delivery promise turned into a delivery wait, and service was reacting instead of planning.
Why PartsCloud
Before the decision, Haver & Boecker deliberately weighed three alternatives:
Four points decided it for PartsCloud
-
AI-based planning instead of a 12-month average
For each part, PartsOS Forecast selects the most accurate of more than 50 forecasting methods and tests it in a backtest against the part's own consumption history over the last twelve months. It detects seasonality, peaks, life cycles and consumption patterns, and can take Haver & Boecker's specific machine knowledge into account. Reactive planning becomes proactive and data-driven.
-
Complements the ERP rather than replacing it
PartsOS Forecast runs alongside the existing SAP system. It replaces nothing and rules out no future ERP decision. Progress in spare parts planning and the big S/4HANA question run in parallel, not one after the other. This keeps effort and IT risk low and delivers results this year.
-
The team keeps the decision
The AI provides the forecast and the priority; the planner still makes the ordering decision. Existing workflows stay in place, only on a better basis.
-
A direct lever on every pain point at once
Service level up, process costs down, scrapping down, and stock built up exactly where it is missing today: demand-driven rather than across the board. The service level is at the center. For Haver & Boecker it is the most important lever and the key target.
The setup: a deliberately small start, expanded step by step
Of around 27,000 items, about 10,000 moved more than five times over the past five years. That is the plannable core. Around 3,000 of these are set to become forecastable, which is where the biggest lever lies.
The project came from the planning team, not from IT, and there was no tender. The business unit got it started.
- March 2026, kick-off: Started from within the planning team, beginning with regularly purchased parts.
- Summer 2026, focus: From calendar week 22, a targeted focus on the items "that hurt us most in the past", meaning the spare parts that were repeatedly late.
- October 2026, first batch: The first forecasts are approved. Transfer to SAP is manual for now.
- From January 2027, integration: Approved forecasts flow directly into the MRP run as planned independent requirements.
Running in parallel:
- Data integration via structured Excel uploads as a pragmatic start before deeper SAP integration
- Regular alignment meetings between the Haver & Boecker and PartsCloud teams to calibrate forecasts and handle special cases
- A fundamental shift in the planning team: away from rigid rules toward service-level-driven stocking. This is deliberately not just a tool topic but a change in mindset that Haver & Boecker is actively driving.
Hannes Malczewski, Haver & Boecker OHG
Day to day: from firefighting to focus
Before
- Every proposal was checked individually in SAP against consumption over recent years.
- That took half a working day per 100 materials.
- Non-critical and conspicuous items were not distinguished; every item got the same level of review.
- The average served as the forecast, and the service level was a by-product.
With PartsOS Forecast
- Non-critical items go into bulk approval automatically.
- The team reviews only the outliers instead of recalculating every proposal.
- A mathematically sound, statistical approach replaces averages plus gut feeling.
- Service comes into sharper focus: working capital goes where it creates availability.
What the Business Case shows
Before the start, PartsCloud and Haver & Boecker calculated a business case together. These figures are a projection for full rollout, not measured results. They will be validated during the ongoing rollout. [⚠ check]
15%
higher availability
< 6 months
payback period including peak phase
up to 1 Mio. €
ROI-amount per year
What's Next
- Validation of the business case assumptions in live operation
- From January 2027: automated handover of approved forecasts as planned independent requirements directly into the MRP run
- 2027: rollout across the full planning range and to further planning teams
- Internal service-level discussion: higher stock at strategic points as a deliberate investment in delivery capability
- Separating spare-part and machine consumption: Many items are needed both as spare parts and for new machines. This has to be cleanly separated in PartsOS pre-planning.
The pace is set. Service business means: deliver. That is exactly what Haver & Boecker aims to do a little better every day.
What other companies can take away
FAQs
-
What does Haver & Boecker use from PartsCloud?
Haver & Boecker uses PartsOS Forecast, PartsCloud's AI-based spare parts forecasting solution. The project started in March 2026, and the first forecasts were approved in October 2026. Expansion toward around 3,000 forecastable items is planned.
-
How does PartsOS integrate with Haver & Boecker's SAP environment?
At the start, via structured Excel uploads: data export from SAP, upload to PartsOS Forecast, and transfer of approved forecasts back to SAP. From January 2027, forecasts flow directly into the MRP run as planned independent requirements.
-
What alternatives did Haver & Boecker evaluate?
SAP S/4HANA as a future option, an in-house solution, and other providers in the spare parts space. PartsCloud won thanks to AI-based planning, complementing the existing SAP without a system change, keeping decisions with the planning team, and a direct lever on all relevant pain points.
-
How fast is time-to-value?
The initial load of nine files takes around 1.5 hours. Haver & Boecker started in March 2026 and approved its first forecasts in October 2026. Scaling and integration happen step by step.
-
Does the AI make ordering decisions?
No. PartsOS Forecast provides the forecast and priority. Non-critical items go into bulk approval, conspicuous ones into individual review. The team approves, adjusts or postpones every forecast before it goes into the ERP.
More Customer Stories
Find out how other industrial companies have transformed their spare parts planning with PartsCloud.
Coperion
How Coperion used PartsOS to virtually eliminate impending stockouts on critical components, increase parts availability by over 15%, and achieve full ROI within just three months.
EUCO Rail AG
How EUCO Rail is transitioning, together with PartsCloud, from reactive, spreadsheet-based planning to proactive, AI-powered parts management, ensuring critical components are reliably available across all locations.