IN Brief:
- Covetrus has gone live with RELEX forecasting and replenishment across 28 US distribution centres.
- The veterinary wholesaler manages more than 10,000 SKUs through the network.
- The deployment targets more automated planning, channel-level forecasting, scenario planning and tighter inventory control.
Covetrus has deployed RELEX forecasting and replenishment technology across its 28 US distribution centres, bringing planning for more than 10,000 SKUs into a common system serving veterinary practices nationwide.
The veterinary wholesaler has gone live with the platform following an implementation supported by Demandtex. The project covers demand forecasting, replenishment and inventory planning as Covetrus seeks to coordinate decisions across sales, finance, operations and inventory.
Forecasting at SKU level becomes difficult when aggregate demand masks substantial differences between products, markets and sales channels. A national wholesaler can experience relatively stable total demand while particular items see local spikes, substitution or seasonal changes.
The resulting forecast is only one part of the planning task. Replenishment logic has to combine expected demand with available inventory, inbound supply, lead times and target service levels before determining what stock should move into each distribution centre.
Across 28 facilities, those decisions interact. Holding more of a product in one location may improve local availability but reduce the amount available elsewhere, while adding stock across the whole network increases working-capital and storage requirements.
RELEX says its AI-driven forecasting and replenishment capabilities are intended to reduce manual planning tasks, improve forecast accuracy at individual-channel level and optimise inventory across Covetrus’ network. The implementation also supports scenario and consensus planning.
Scenario planning allows teams to examine how alternative demand assumptions affect inventory and replenishment requirements before those assumptions become purchase orders or stock transfers. Consensus planning provides a structure for sales, finance and operations to reconcile their expectations rather than maintaining separate forecasts.
That process matters because each function can view the same inventory differently. Sales may focus on product availability, finance on working capital and operations on capacity inside individual distribution centres. A shared planning system cannot eliminate those competing priorities, but it can give them a common demand and inventory baseline.
Covetrus sought to expand its planning capability as its distribution network scaled. RELEX states that the project is intended to streamline planning, reduce inventory days of supply and improve collaboration across operations.
The 6 October announcement says Covetrus has gone live with the technology, but it does not provide quantified before-and-after figures for forecast accuracy, inventory days, service level or labour savings. Those points should therefore be treated as implementation objectives and software capabilities rather than proven results from the completed deployment.
That distinction is particularly relevant for AI-labelled planning technology. Forecasting performance depends on the quality of historical demand data, product records, lead-time information and the way planners manage exceptions. An algorithm cannot correct unreliable source data simply by processing it at greater scale.
A common platform can nevertheless reduce some of the manual work created by a fragmented planning process. Automating routine forecast and replenishment calculations allows planners to concentrate more time on exceptional demand, supply interruptions or products whose behaviour does not fit the standard model.
Veterinary distribution also combines high-volume routine products with specialist items that may have very different demand patterns. Applying one inventory policy across more than 10,000 SKUs would therefore create unnecessary stock in some categories while risking shortages in others.
Planning systems can segment products and apply different replenishment rules, but those policies still have to reflect actual service requirements. Fast-moving lines may justify frequent replenishment and tighter statistical forecasting, while slower or more specialist items require different safety-stock and ordering assumptions.
Demandtex supported Covetrus and RELEX during implementation. Large planning deployments typically require data preparation, process mapping, integration and user adoption alongside configuration of the forecasting application itself, making implementation work part of the operating change rather than a separate technical exercise.
The go-live moves Covetrus into the stage where its assumptions can be tested against live order behaviour. Forecasts can now be compared with actual demand, inventory levels can be monitored against service requirements and replenishment policies can be adjusted as performance data accumulates.
The useful measures will therefore be operational rather than technological: whether product availability can be maintained with lower excess inventory, whether routine planning requires less manual intervention and whether sales, finance and operations work from more consistent assumptions when demand changes.
RELEX has supplied the forecasting and replenishment platform across all 28 US distribution centres. The next evidence required is measured performance from that live network, which will determine whether the planning model produces the inventory and forecasting improvements set out in the implementation objectives.
