Good companies plan—great companies execute the plan


Lots of vendors are doing supply-chain planning and transportation management. Unfortunately, these activities are disconnected, and that disconnect is costly.

Making a plan that can’t be executed

Planners use systems that react to demand fluctuations, but these same tools don’t have the capability to create replenishment plans that fully meet the needs of the supply chain. For example, they don’t understand that:

  • The warehouse in Los Angeles is nearly full and can’t accept more than a few loads
  • Sites are constrained on the number of shipments they can make or receive. Labor doesn’t grow on trees.
  • Carriers don’t have infinite capacity. Generally, adding volume forces new carriers into the mix, raising costs or degrading quality.

So, plans get developed that can’t be executed economically or without superhuman effort, and faced with this unconstrained plan, warehouse, and transportation managers battle to execute. Failure to find a carrier for a load or a warehouse to not unload a shipment can generate a customer service failure for which they are blamed.

The cost of a bad plan

Executing an unconstrained plan can often require the investment in extra labor either on a spot basis or, worse, overstaffing to be ready for these peaks. Similarly, there is a real cost for hiring carriers not on the preferred list, potentially in both dollars and service. For example, when volume peaked, a large CPG shipper was forced to move away from cheaper intermodal. This forced a move to expensive truck shipments.

There is a better way:  create a plan that understands the various constraints and prioritizes what needs to be deployed. This plan may only peak, i.e., push for large expenditures if and when absolutely needed to maintain high service levels.

Creating a plan that can be executed

Creating a plan that can be executed requires significant data and the capability to manage all the interactions in the entire network simultaneously. This is because reducing the volume out of one site because it exceeds its throughput capacity will impact all the sites that it supplies. Increasing volume into one site may force the reduction of shipments into another. All these interdependencies need to be considered to create a holistic plan.

Taking the data from an existing replenishment system and adjusting it to meet the capacity limitations in the network is not a trivial task. This is further compounded by the need to consider, for example, in CPG networks, the moves between sites are done in full truckload increments. So when it comes to prioritizing what gets pushed into a warehouse, for example, space-constrained, it is necessary to understand the priority of the entire shipment. This prioritization, therefore, requires to know what is on each vehicle.

Adding to this complexity is the need to look out day-by-day for multiple weeks to manage such things as month-end demand spikes. Seeing a significant increase in demand in three weeks gives the needed time to progressively work to pull forward high-priority deployments and push back the less-needed product.

LevelLoad creates a cost-feasible, high-service plan

Provision AI has developed software called LevelLoad. LevelLoad uses advanced linear programming combined with an ultra-fast load builder. This load builder leverages reinforcement learning, a relatively new form of artificial intelligence, to establish what goes in each vehicle. Using this patented technology, a solution can quickly be created. The process is:

  1. Analyze: LevelLoad takes in the output of the advanced planning system (SAP APO, Kinaxis RapidResponse, BY,  OMP, etc.) along with all of the constraints in the network (shipping capacities, carrier rates, item master data, and facility throughput capacity)
  2. Smooth: LevelLoad performs an overall optimization to “smooth” shipments across the network and reduce volatility over the next ~30-60 days. This smoothing gives detailed information on how many transportation units (truck, intermodal, etc.) must be sent on every lane daily for the next 30-60 days.
  3. Tender: LevelLoad integrates with existing ERP and TMS solutions to automatically tender freight much further in advance than a traditional planning team would allow
  4. Load: Once the ship day arrives, ProvisionAI’s other tool, AutoO2, looks at the latest demand signal to make sure that EXACTLY the right, high-priority products are put on a replenishment order, optimizing load building completely for weight and cube
The benefits of creating and executing a great plan

The benefits are enormous. In addition to eliminating the need for superhuman effort to adjust infeasible plans manually, get additional trucks, or push through or store way more volume than the warehouse should, there are substantial financial savings as:

  • The carriers you want accept the tenders.
  • The same carriers recognize that the reduction in volatility has a significant financial benefit for them, so they offer better rates for lanes with less day-to-day volatility.
  • Warehouses are less taxed. Customers receive the inventory they need because it has been prioritized, and because the site has capacity, it’s unloaded in time to meet their order.

This is not theoretical. A large consumer products company has already attested to the millions they save yearly. The savings are not the only benefit; customer service has improved in the form of on-time and in full. They couldn’t have done this without having a tool that transforms the excellent plan from their supply planning system into a great one.

Good companies plan. Great companies execute. They do both.