Congratulations to Kimberly-Clark and LevelLoad
Says Kimberly-Clark, upon receiving this award, “So we talk a lot about how to speed up the supply chain, and with the combination of Rapid Response and ProvisionAI, we did a proof of concept in February of 2021 and implemented the solution in October 2021. We now know how to plan how our products move around the network and how to prioritize inventory replenishment, saving us millions of dollars in transportation costs. This has been a cost game changer for our North American supply chain.”
We are proud to present an innovational deployment transportation scheduler LevelLoad, that cuts transportation costs by eliminating volatility. Filling the gap between supply-planning and execution systems, LevelLoad is an essential component of supply-chain excellence. LevelLoad from ProvisionalAi uses Automatic Order Optimization (AutoO₂) load building optimization to create legal and efficient shipments that fill more trailers.
Kimberly-Clark fully deployed the platform (LevelLoad) across all North American operations; as a result of both the process improvements and new technology, it reduced variability daily by 40%, particularly in locations where production plants are shipping to its distribution centers.
Deployment volatility and the associated high costs
Fluctuations in volume (volatility) have always been a part of all supply-chain networks. Customer orders are volatile, and that drives volatility in inventory deployment. Moreover, the replenishment movements of products from plants or suppliers to market-facing distribution centers could be more balanced. The following paragraphs outline the cost impacts and effective ways to mitigate them.
Deployment volatility generates excess cost
If only everything were calm and smooth – but it isn’t. Instead, deployment volumes on any lane (origin-destination-mode) vary significantly day-to-day. It generates a considerable number of bad outcomes:
• Best (preferred) carriers may be unable to handle peak volume, so some shipments may need to be shifted to higher-cost, less reliable truckers.
• Both shipping and receiving locations may become inundated with volume and unable to ship or receive without substantial overtime labor. It is worse in the receiving location because a tsunami of volume from several uncoordinated-ship points arrives simultaneously.
Drowning in this volume can generate detention and potentially strand needed inventory in a queue of trucks waiting to unload. The impact on a critical measure called OTIF (on-time, In full) can create significant chargebacks. Constraining deployment alone to match capacity is insufficient in customer-facing distribution sites. In these sites, customer volume typically takes precedence, and it, too, must be accounted for when trying to constrain site throughput to match capacity.
• While the dock is a constraint on throughput, the lack of storage space can also adversely impact unloading. Often this generates significant service failures as much-needed product waits in trailers and trailers wait to unload.
Supply-planning systems may exacerbate supply-chain volatility
Supply-planning systems typically react to customer needs unpredictably, creating much supply-chain volatility. That is, the volatility in the number of deployment stock transfer orders (STOs) each day is a reaction of the supply-planning system to customer orders and changes in safety stock generated to mitigate this volatility. Furthermore, because supply-planning systems have limited capability to consider transportation costs and preferred carrier availability over time holistically, the supply signal tends to be problematic. So supply panning can create self-inflicted wounds.
Additionally, we know of no supply planning systems that actively manage site volume throughput. Managing the throughput of receiving locations is particularly complex because receiving volume comprises shipments with varying transit times based on shipping location and mode. In summary, most (all) planning systems don’t manage volatility and throughput, and their reactions to customer variability may amplify volatility. When you add to these shortcomings the need for more understanding of site-space constraints, the picture develops of a significant and expensive problem caused by the gap between supply-chain planning systems and execution.
Filling the gap between supply chain planning and execution
There are multiple ways to impact volatility and its related costs:
- Smooth freight volumes over time by pushing lower-priority deployment demand into a later date and pulling forward more needed requirements
- Cut the pressure on shipping and receiving sites by shifting limited capacity between lanes
- Increase the lead time between awarding loads to preferred carriers and the shipment date. Additional notice means available carrier capacity can be secured.
Under all circumstances, there is a need to maintain customer service. However, it requires a careful balance between inventory availability and cost. There is no sense in saving a dollar and hurting a customer. Hence, it is crucial to prioritize using the scarce capacity to move the most needed products.
Smooth freight volumes over time by pushing and pulling when deployment happens
Supply-planning systems typically project when the product is needed based on target stock levels, customer demand forecasts, and variability factors for demand uncertainty and transit time.
If there is an effective “trading-off” of risk and reward, bringing high-priority demand earlier or pushing items with ample stocks later does little to impact service. This strategy can effectively eliminate deployment peaks well before they encounter.
Deployment Transportation Scheduler smoothing volatility
Cut site volatility by shifting limited capacity between lanes
As mentioned earlier, managing site throughput and storage capacity is equally crucial to managing lane volumes. Generally speaking, when making daily deployment decisions, the requirements on any lane range from the high-priority, most-needed demand to lower-priority needs. And some deployment lanes will likely require many more high-priority replenishments than others. Therefore, replenishment transportation scheduling systems need to use the preponderance of priority requirements to favor volume on one lane over another on any given day while setting the total site volume to within capability limits.
Reduce the amount of higher-cost freight by tendering earlier
Tendering deployment loads early (as many companies do) can create problems if it ties the tendering of trucks to the definition of what is on each load. The problem is that the further in advance you determine what items to replenish, the less likely it is to be correct. But, because, for the most part, carriers don’t need to know what is on the load, it’s possible to reserve trucking capacity without specifying what will go inside the trailer. In addition, the determination “later” allows the planning system to have more information in the form of orders and inventories, thus enabling a more accurate determination of the priority of need at any customer-facing distribution site.
This postponement strategy generates the following results:
• Increased lead time for the carrier because they still receive the tenders early and “first tender acceptance,” – meaning the “most favored carrier” takes the load.
• Better tailored of actual stock-transfer orders (STOs) because the contents are specified as late as possible.
LevelLoad and AutoO₂ bridge the gap between planning and execution to cut costs while maintaining service
LevelLoad, a SaaS-based replenishment transportation scheduler, has proven effective in smoothing deployment volatility. Together with AutoO₂, an optimizing load builder, they can perform the needed smoothing and postponement.
Here is how it works
When the supply planning system runs, it sends the latest requirements and other data, like forecasts and production, to LevelLoad. It is augmented with data from the ERP of existing orders, inventory, and previously positioned reservations of carrier capacity.
It enables LevelLoad to generate a plan for a period of time, usually, ~30 days, constrained by site space and throughput capacity while maximizing the deployment of high-priority deployments.
LevelLoad uses fast-linear programming techniques combined with an ultrafast AI-enabled (reinforcement learning) load builder that simulates the contents of future loads. LevelLoad iterates between the linear program and the load builder to identify what trucks need to schedule to move which products are on which lanes.
It quickly settles on a plan even for large networks and creates the required signal for the TMS (Transportation Management System) to perform early tendering. It is often done at night and can be completely autonomous.
Deployment Transportation Scheduler LevelLoad uses mathematical optimization and AI.
Closer to the shipment date, generally, the day before the load shipment, AutoO₂ (an optimizing load builder that considers many more factors than LevelLoad) takes the latest requirements information from the supply planning system. With this data and information about how many loads have been scheduled by the LevelLoad process run days earlier, AutoO₂ determines what should ship on each vehicle.
Thus, each of the tenders initiated by LevelLoad, are filled with the highest-priority, most-needed product. The diagram of the process timing follows.
Using load builder AutoO₂, which loads all ships ultimately – no wasted capacity in either cube or weight.
• Because the deployment transportation scheduler LevelLoad knows the carrier in advance, loads tailor to utilize that specific carrier’s equipment. For example, AutoO2 uses a carrier with lightweight trucks and provides instructions to load it heavier than a “standard” hauler.
• Importantly, AutoO₂ shows the placement of each pallet on the load and, if needed, each picked case. This way, the shipping site can be sure the load will be legal and arrive damage-free.
The proof is always in the pudding. We can just share the results if you get in touch with us.