Find your surplus-to-risk pairs

Network visibility surfaces transferable stock before you cut new POs.

Talk to an MRO expert →

Short answer: Multi-site manufacturers reduce working capital by $20M on average per company by applying AI-driven optimization across all distributed locations simultaneously, using criticality and failure rates instead of demand-forecast models to set stocking levels, based on Verusen customer results. The approach works across multiple ERPs without requiring data cleanup first because it ingests inventory as-is and surfaces redundancy, dead stock, and stockout risk across all sites in one view. A Fortune 500 CPG manufacturer with 41 sites identified $63M and verified $60M in MRO savings while cutting material review time from over 20 minutes to 4 minutes per SKU.

Spare parts inventory optimization: The practice of right-sizing on-hand inventory for non-moving maintenance, repair, and operations (MRO) parts across one or more manufacturing facilities to minimize working capital tied up in excess stock while reducing stockout risk on critical items. Unlike demand planning for finished goods, MRO optimization uses failure criticality, asset downtime cost, and lead time to calculate safety stock, since spare parts fail unpredictably rather than sell on a schedule.

Why multi-site spare parts inventory behaves differently

Multi-site manufacturers typically carry 20 to 30% excess maintenance, repair and operations inventory while simultaneously facing stockout risk on 10 to 15% of critical parts, industry estimates suggest, consistent with Verusen's experience across hundreds of implementations. The root cause is not data quality: it is fragmentation. Each plant operates its own ERP instance or manual tracking, so a maintenance engineer at one site cannot see what is on the shelf at another location, and procurement cannot cross-reference usage patterns across the network.

When a bearing fails at Plant A, the team orders a replacement unaware that Plant C is sitting on units that have not moved in years. When the same bearing fails at Plant B three months later, they order again. The result: parts accumulate where they are least needed and vanish where they matter most. Aggregating inventory counts across sites does not solve this because a part with 100 units at Plant A and 50 units at Plant B does not mean you have 150 usable units. If the part fails unpredictably and a failure at Plant B stops the production line, inventory at Plant A is worthless.

Three decisions that govern multi-site stock levels

Location-specific criticality, not total count, determines adequate stock. Multi-site optimization requires three simultaneous decisions applied to every SKU across your network.

1. Failure likelihood at each location: Does this part fail predictably, or only when demand spikes? A bearing that fails twice in five years at Plant A may fail weekly at Plant B because of age, environment, or operating tempo. Safety stock rules that work for Plant A create waste at Plant B.
2. Cost of a stockout at that specific site: A stockout on a redundant pump costs zero; the same stockout on a single point of failure stops the line. Calculate the per-hour downtime cost, equipment damage risk, and production loss at each location. That number dictates your target service level there.
3. Network routing feasibility: Can a part reach Plant B from a hub in 24 hours, or does distance require stock on-site? Lead time, geography, and criticality together determine whether pooling works or whether you need location-specific safety stock.

A Fortune 500 CPG manufacturer applied all three frameworks to 41 sites and cut material review time from over 20 minutes to 4 minutes per SKU, based on Verusen customer results. Georgia Pacific, operating 110 US sites across four ERP systems, used this framework to centralize stocking decisions from hundreds of decentralized engineers to a core team of 7, recovering 6,600 hours of manual review and flagging 2,900 materials at stockout risk across the network, based on Verusen customer results.

The hidden cost of siloed, site-by-site stocking

When each site manages spare parts independently, you lose visibility to move overstock to shortage sites, and material review time balloons: a Fortune 500 CPG manufacturer across 41 sites reduced review time from over 20 minutes to 4 minutes by consolidating inventory visibility into a single optimization layer (based on Verusen customer results).

Siloed stocking creates two simultaneous failures. First, excess inventory accumulates at sites where certain assets rarely fail, tying up working capital and driving obsolescence. Second, sites facing genuine demand encounter stockout risk because no one sees the total inventory picture across the network. You end up managing dozens of independent stocking policies instead of one coherent strategy.

The visibility gap: why site-level data does not scale to multi-site decisions

Each site's ERP or EAM holds accurate local inventory and maintenance history, but that data does not connect to the plant 50 miles away. When a bearing fails at Site B, the maintenance team orders it because they cannot see Site A holds twelve identical bearings in storage. Multiply this across 20, 50, or 100+ sites, and the fragmentation becomes structural: parts scatter across the network, no one sees the total picture, and capital efficiency collapses.

How to move from site-by-site decisions to centralized network governance

The shift from fragmented to unified stocking requires a clear governance layer. Georgia Pacific centralized decisioning from hundreds of site-by-site personnel to a team of 7 by implementing a centralized materials authority that owns three types of decisions (based on Verusen customer results):

1. Local decision: site engineers retain authority over parts unique to their asset fleet or with lead times under 3 days; these stay in site inventory.
2. Escalation decision: parts flagged as dual-site or multi-site inventory go to the materials authority for approval; authority compares criticality, lead time, and failure frequency across all sites before approving purchase or reallocation.
3. Network decision: slow-moving or high-cost spares are consolidated into a central hub and allocated on demand, reducing total on-hand inventory while preserving uptime.

This governance model replaced hundreds of manual spreadsheets and phone calls with one source of truth. The CPG manufacturer reduced material review time to 4 minutes per part because engineers could now see full network inventory and criticality in real time, freeing the team to focus on strategic reallocation rather than reactive firefighting.

Setting network-wide min/max rules by criticality and lead time

Once you have network visibility, replace site-by-site safety stock guessing with a single decision rule tied to criticality and lead time:

Criticality- and risk-based stocking across a network

Criticality-based stocking eliminates the inventory mismatch that demand formulas create: a bearing that fails once every 18 months on a revenue line gets zero stock under a sales-forecast model, but stops production for three weeks when it inevitably fails. Seadrill, a global offshore operator with 17 rigs, applied criticality-and-risk-based stocking across its fleet and shifted from redundant safety stock on parts that failed once per decade to adequate stock on the ones that stopped operations every 18 months, then enabled hub-and-spoke shorebase-to-rig distribution based on Verusen customer results.

Why demand formulas fail for spare parts

A maintenance engineer at a food processing plant had a bearing that failed once every 18 months on a revenue-critical line. The plant carried no stock. When the bearing failed, the line stopped for three weeks. SAP's demand formula returned zero not because the part was unimportant, but because there was no formal purchase history. Formulas designed for finished goods cannot work when failure patterns are irregular and the part has never been bought on a schedule.

Building your criticality-first classification

Define criticality by production consequence, not part cost. A critical part is one whose failure stops the line or reduces output on a revenue-generating asset. Create three tiers:

• Line-stop parts: failure halts production entirely.
• Degraded-mode parts: failure reduces output or quality but production continues.
• Non-essential parts: failure triggers a scheduled repair with no immediate operational impact.

Map every material in your ERP to one tier using input from maintenance and operations teams. For a 10-site operation with 20,000 active materials, this takes one week. Then overlay failure frequency from your EAM or maintenance logs: parts that failed three times in the past two years go into the high-frequency bucket; parts that failed once in five years go into low-frequency; parts with no failure history belong in low-frequency unless your operators know otherwise.

Setting network-wide min/max by criticality tier

Assign a min/max formula to each criticality tier, then apply site-by-site based on local lead time and failure likelihood. A Fortune 500 beverage producer applied min-max rules tied to criticality across 130+ plants and recovered $35M in North America within phase one based on Verusen customer results.

Pooling and redistributing spares between plants

Multi-site manufacturers unlock 10 to 20 percent of total MRO working capital by pooling spare parts strategically across plants instead of stocking every part at every location, based on Verusen customer results. A major US energy company managing 45,000 materials across multiple facilities achieved $29.7M in verified inventory reductions by consolidating stocking authority and redistributing stock based on real criticality rather than historical hoarding, while maintaining uptime through hub-and-spoke coordination.

Hub-and-spoke decision matrix

The pooling decision turns on three variables: criticality (how fast the part fails), lead time (how long to get a replacement), and failure frequency. Map each material to this matrix to determine where it lives and how it is restocked. Once classified, your ERP or inventory system must track transfers between sites and alert the hub when distributed stock falls below its minimum threshold, so satellite plants never wait for a critical part.

AI-driven rebalancing and dispatch

Pooling only works if you can see surplus and deficit in real time across all sites. Purpose-built inventory optimization platforms ingest your ERP and EAM data as-is and return a prioritized rebalancing list ranked by capital unlock and execution risk. For example, a platform ingesting SAP across 45 plants surfaces materials sitting idle at Plant A while Plant B is one failure away from stockout, then recommends which transfers happen first and which can wait 30 days without production risk. A Fortune 500 CPG manufacturer applied this framework to 41 sites and cut material review time from over 20 minutes to 4 minutes per SKU, eliminating the manual hunt across multiple systems that causes plants to hoard inventory.

1. Run one full inventory audit across all plants using the criticality and lead-time matrix above.
2. Flag any part where current on-hand exceeds the hub's recommended stock level by more than 50 percent; these are rebalancing candidates.
3. Transfer surplus from overstocked plants to the hub or to plants facing stockout risk within 30 days.
4. Update ERP transfer and minimum-threshold alerts so the hub can redistribute without manual intervention on future cycles.

See the ranked transfer list

AI ranks every redistribution by savings and risk, live on your data.

Talk to an MRO expert →

Setting network-wide min/max and safety stock

Safety stock for spare parts must be set by criticality and lead time, not demand frequency, because a part that fails twice in five years has no demand history for formulas to use. A leading gold mining company with 17 sites across three ERP systems applied this logic and identified $96.8M in MRO inventory savings (based on Verusen customer results).

Why demand-based formulas fail for spare parts

Standard safety stock formulas require consumption data. When a bearing fails twice in five years, that data does not exist, so the formula returns zero. The plant carries zero stock. Then the bearing fails and the production line stops for three weeks. Across your network, the parts you overstock never fail, and the critical ones you need, you do not have.

The decision rule: map criticality and lead time to min/max

Each part needs a stocking policy tied to two variables: consequence of failure (critical, semi-critical, non-critical) and supplier lead time. Longer lead times and higher criticality require larger buffers to absorb delays and demand spikes.

How to apply this across multiple sites

1. Classify each part by consequence of failure (critical, semi-critical, non-critical) — does it stop production, slow it, or affect neither.
2. Record the supplier lead time and count of approved sources for each part.
3. Map the part to the criticality + lead-time band in the table above; apply that min/max rule network-wide.
4. Configure each site's ERP min/max fields using the calculated values; override only if local sourcing changes the lead time.
5. Review and rebalance quarterly as sourcing or production risk changes.

The same bearing may have different policies across three plants because lead time and criticality are local, not universal. When you apply this logic, you eliminate the false choice between overstock and stockout, and recover the 20 to 30% of working capital locked in excess inventory (industry estimates consistent with Verusen's experience across hundreds of implementations).

AI automation step. Once you define min/max by criticality and lead-time band, AI-powered inventory systems rank rebalancing opportunities by capital unlock and execution risk, then surface a prioritized action list to your materials team, so you move from manual spreadsheet reviews to a ranked decision log in minutes instead of hours.

How AI optimizes spare parts across every site

AI optimizes spare parts across multiple sites by ingesting inventory, demand, criticality, and failure data from every ERP and EAM system simultaneously, then calculating the right stock level for each part at each location based on actual failure patterns rather than demand forecasts. The platform ingests 41M+ unique MRO materials from hundreds of implementations and maps each to a criticality-lead-time decision cell, returning a rebalancing instruction list your team can execute in 48 hours.

Unlike demand planning tools built for finished goods, AI-native inventory optimization recognizes that spare parts don't sell on a schedule: they fail unpredictably, and the cost of a stockout far exceeds the cost of holding the part. Industry estimates suggest the average asset-intensive manufacturer carries 20 to 30% excess MRO inventory and simultaneously faces stockout risk on 10 to 15% of critical parts, consistent with Verusen's experience across hundreds of implementations. The method that fixes both problems rests on three concrete decision rules.

The three decision rules: criticality, lead time, and consequence

1. Criticality first: A bearing that stops your production line ranks higher than a bolt with ten backups. The AI assigns a criticality score based on which assets depend on the part and downtime consequence if it fails.
2. Lead time second: A part with a 12-week lead time needs more buffer stock at your primary site than one available in 48 hours. The algorithm weights stock levels by supplier lead time and distance from each location.
3. Consequence of stockout third: The AI calculates the right quantity and location for each part based on failure frequency, lead time, and the cost impact of a stockout, then surfaces a materials-rebalancing list ranked by working capital unlock and execution risk.

A Fortune 500 CPG manufacturer applied all three frameworks across 41 sites and cut material review time from over 20 minutes to 4 minutes per SKU. A major US energy company identified $40M in excess inventory across 45,000 materials reviewed in under a year and verified $29.7M, achieving 100% audit capability for FERC compliance, based on Verusen customer results. When criticality and lead time drive your allocation, a part that has failed three times in the past year at Site A but never at Site B gets stocked differently at each location, even though both sites own the same equipment.

The outcome: working capital unlocked, stockout risk eliminated, and stocking policy changes distributed to your team in a single, executable list. No manual recalculation. No guesswork.

Go deeper: this article supports our pillar guide, MRO Inventory Optimization: The Complete Guide. Related: MRO inventory strategy for multi-site manufacturers.

Further reading: spare parts inventory management guide, MRO spares inventory optimization guide, and MRO inventory optimization best practices.

Pool one part family first

Start with bearings or seals and verify the working-capital release.

Talk to an MRO expert →

Frequently asked questions