Most of these pain points can be summarized in two broad issues:
1. Inventory projection
- Out-of-stock scenarios: Inventory not available to sell either due to under projection or incorrect timing (e.g., misalignment with demand)
- Inventory overage: Inventory procured more than what’s needed to satisfy the demand
2. Inventory placement
- Split shipments: Orders with more than one item could potentially ship from more than one fulfillment center
- Longer delivery times: The longer the distance between the customer ship-to location and the inventory location, the longer it takes to deliver the order to the customer
- Elevated shipping costs: Orders shipping from a farther fulfillment location result in higher shipping cost
- Higher inventory carrying cost: Inventory spread across a wider store network or placed in sub-optimal stores, resulting in higher inventory carrying cost
Addressing inventory projection issues
The key to successful inventory projection is a highly accurate demand forecasting process that uses past sales history, future growth, and other internal and external factors. These are a few factors to consider while implementing your demand forecasting solution:
Flexibility & Dynamism
Forecasting tools should be flexible enough to accept a wide range of input parameters. Sales history and future growth parameters are fundamental to the forecasting process. It’s a good idea to have at least two years of sales history for a meaningful demand projection. New product launches and new store launches must inherit the sales history from similar products and stores, respectively.
Future sales promotions, campaigns, discounts, and special events (concerts, sports events, tax-free weekend etc.) generally result in sudden spikes in demand and must be planned to make them successful. Additionally, seasonality, geographies, and location insights play an important role along with factors like economic and political circumstances – e.g., pandemic or support for green energy initiatives.
Real-time unplanned events (both internal and external) can have positive or negative impact on sales. A good forecasting tool should help your business to quickly react to these events and minimize the impact or maximize the revenue. This is possible if you can quickly adjust the input parameters, customize the models and features in real time, generate and regenerate the forecast as needed.
Accuracy
Demand forecasts are generally uncertain and inaccurate to start with. The key is improving the accuracy over time until the error rate falls within an acceptable threshold. Businesses can achieve this through benchmarking, performance measurement, ongoing optimization, and accurately interpreting sudden spikes in sales and dynamic scenarios due to unexpected events.
Insights and reports
Analytic insights are what fuel strategic business decisions about fulfillment. Several factors could be driving the future demand up or down. Some factors are significant and others, not so much. Timely and actionable insights into the drivers of these future demand fluctuations will help you proactively react by adjusting future inventory levels.
Business intelligence dashboards and reports allows users to foresee future out-of-stock scenarios and replenish the inventory quickly. Visualization of projected demand and supply levels via charts, tables, and gauges using interactive dashboards is key in maintaining optimum inventory levels. Especially when you can drill down to different levels of granularity - e.g., projection by region or by DC/store, by time, by fulfillment type (BOPIS, ship-from-store), by channel and by product hierarchy (division, department, class, sub-class, category, or SKU).
Scalability
As companies grow or diversify, the need for extending into additional regions, channels, markets, or product lines with minimal effort in configuring the forecasting process goes a long way and helps with speed to market.
Scalability to perform at a global level for those retailers looking to expand internationally may require additional considerations like lead times, local laws around product restrictions (e.g., country of origin), and local shopping patterns.
Artificial intelligence (AI) and machine learning (ML)
Technology helps retailers to understand customer behaviors and preferences and respond quickly and accordingly. Data-driven AI/ML-based forecasting has gained a lot of traction in the recent years.
This approach involves self-learning models that incrementally improve forecast accuracy based on feedback provided. One good example is consideration for real-time external factors such as commodity prices, income indicators, market performance, social media trends, etc. Machine learning algorithms can predict changes in consumer demand more precisely and can quickly recognize unusual demand patterns, complicated relationships in large datasets, and capture signals for demand fluctuation.
Predictive analytics provide insights into why a consumer is buying a product. This offers a deeper understanding of customer preferences and thereby gaining an edge over the competitors.
Addressing inventory placement issues
Retailers who have a significant physical store presence have expanded their fulfillment capabilities outside the traditional approach of shipping from a distribution center, by offering buy online pick up in store (BOPIS), ship from store (SFS) and ship to store fulfillment options. Many retailers are also exploring curbside pickup and same-day delivery.
These approaches expose the inventory across the distribution center and store network to online sales and the positioning of this inventory becomes extremely critical, considering increased customer expectations in the recent years.
So, the question keeping retail leadership up at night is how to bring the product closer to the customer so that his/her order can be fulfilled within a set of parameters that helps to keep the customers happy without hurting on the margins.
