8 keys to ecommerce modeling success

Consumer behavior continues to shift. It’s increasing the demand for distribution network modeling of ecommerce supply chains. There are notable differences when modeling networks that supply retail versus those that support ecommerce. Here are eight crucial points for successful ecommerce modeling.

1. Model in units of measure consistent with ecommerce demand.

Distribution networks are often modeled in units of measure that readily support transportation or warehouse costing (for example, units, pallets, cases or weight). However, using such traditional units of measurement makes it challenging to apply fulfillment capacity constraints. It’s also difficult to represent the impact of creating split shipments due to different deployment strategies and consumer ordering behavior.

It’s essential to create a distribution network that meets real-world ecommerce demand and builds decision-makers’ confidence in proposed solutions. Doing so requires representing demand in the same units of measure that increase usage and cost (for example, orders, order lines or shipments).

2. Develop an understanding of daily ecommerce demand variability and the mix of order line types.

Distribution networks are often modeled as single-period models. This is due to the inherent complexity of representing the many flow-paths of products as they move through the network. In contrast, the complexity of modeling ecommerce fulfillment isn’t typically rooted in the number of flow paths products can follow through the distribution network. Instead, the complexity is more likely found in the various types of fulfillment nodes required to serve lumpy daily demand and the mix of ecommerce order lines and omnichannel order lines.

Representing the lumpiness and mix of demand is vital to creating a distribution network that meets real-world ecommerce demand and boosts decision-makers’ confidence in the proposed solutions.

3. Articulate the difference between time-to-customer and click-to-delivery time.

Typically, network models only measure transit time to the customer. However, shippers are usually more interested in measuring click-to-delivery time because that’s the waiting time ecommerce consumers experience. Transit time and the time of day the order is placed influence click-to-delivery time. So do how long it takes to fulfill the order and when is the order processed for fulfillment and shipping (order cut time).

Articulating the relationship between transit time and click-to-delivery time supports more informed decision-making. It precisely describes the cost and service trade-offs in terms that often make the most sense to decision makers.

4. Use accurate transit times and be aware of potential conflicts between delivery services and desired transit time requirements.

It’s not uncommon to use distance as a proxy for transit time in distribution network models. This is particularly true when most freight moves by full truckload.

However, it’s important to use actual transit times for ecommerce modeling. It supports informed decision making and increases decision makers’ confidence in proposed solutions. Delivery services like the U.S. Postal Service’s hybrid services, UPS SurePost and FedEx SmartPost offer attractive rates but lack consistent transit times. Generally, you shouldn’t consider these types of services if your desired transit times are three days or less.

5. Consider actual shipment weight as opposed to an assumed or average shipment weight.

Often, package weights are such that a shipper pays comparable rates to ship packages over noticeably different distances. As a result, service requirements and facility capacity may be a more significant determinant of deployment than transportation costs. Guaranteeing that transportation costs are accurate improves accuracy and better informs decision making.

6. Determine whether you can fulfill ecommerce orders from both distribution centers and retail stores. If so, it’s vital to have an accurate landed cost for each store.

Developing the cost to land products at each store usually involves a level of detail that’s typically out of the scope of ecommerce-only projects. Often the scope of an ecommerce project assumes that it’s only important to understand the landed cost to each distribution center.

Allowing enough time and resources to develop an accurate landed cost by store supports more informed decision making. It clearly describes the cost and service trade-offs of using all possible ecommerce fulfillment nodes.

7. Model products or orders that incur additional distribution handling costs and capacity use requirements differently.

A single set of distribution handling cost and capacity use assumptions are often used to model distribution networks because all products will be deployed to every distribution node. Therefore, all products within the same distribution location often incur a common handling cost and capacity usage.

In contrast, for ecommerce networks, it’s often important to represent more discreet handling costs and capacity usage rates. One of the main questions for an ecommerce network analysis isn’t simply, “how many distribution centers should we have and where?” You should be asking, “given the distribution centers that we have, which products should be deployed at these distribution centers?”

8. Develop an understanding of which markets represent lucrative opportunities from a demand density perspective.

Understanding which markets you can serve from existing fulfillment nodes — and the overall order density throughout the country — is vital to facilitating candidate site selection. It also focuses the analysis on markets with the most significant opportunity for improvement.

— By Jeff Zoroya