In the retail landscape, demand forecasting is one of the most critical foundations for accurate, timely, and effective inventory management. Most retailers and distributors struggle to have the right product at the right place, in the right quantity and at the right time to meet the demands of their customers. Inventory management is typically the single biggest capital investment a company makes, with the majority of a retailer’s working capital tied up in its inventory investment.
One of the biggest challenges associated with demand forecasting for business executives today is understanding the impact of demand volatility on demand forecasting. To address this challenge, companies often add excess inventory to protect against inaccurate forecasts. Various studies have shown that excess inventory holding costs due to storage, obsolescence, spoilage, and taxes could increase costs by 20 percent to 30 percent.
With the advent of advanced AI and Machine Learning models, retail companies are increasingly using demand forecasting to optimize inventory. With democratization of AI and Machine learning even small and medium sized businesses can leverage powerful AI/ ML technologies to create accurate forecasting models. With improved forecasting, companies can drive better decisions regarding cash flow, risk management, capacity planning, and workforce planning.
Factors influencing retail demand forecasting
Several factors complicate demand forecasting in the retail industry. These include:
Store location: For accurate forecasting it is very important that every location has its own set of SKUs to account for demographic influences on the demand. This gives a more accurate picture which could be later aggregated at higher level to get a macro level view. As the number of products or locations increases the SKU count increases exponentially and this is very AI based demand forecasting plays a critical role.
Changing trends: Retail trends change rapidly. What is a hot trend in one day becomes obsolete the next day. So external factors like trend data influence forecasting to a large extent which needs to be factored in.
Promotional/Causal demand: Consumer demand will be impacted by factors like in store promotions, competitors going out of business and various other factors that fall outside the typical pattern that could be addressed by a purely statistics based forecasting model. This is very advanced AI based techniques like anomaly detection could be applied to factor in these anomalies.
Seasonal or cyclical demand: Majority of retail products have cyclical demand. These could be seasonal in nature for consumer products or driven by a fiscal calendar for business goods. In some cases seasonality impacts the same product/SKU at different locations differently. It is important to forecast factors in cyclical nature at store, region and national levels.
Demand forecasting trends in Retail
Cross functional forecasting
There is a trend towards a holistic perspective on demand forecasting. The demand forecasting function at many companies is still too silo’d with sales and operations using their own data, insights and forecasts. However, many organizations are realizing the benefits of an integrated approach leading to more accurate forecasting and cost savings.
Out growing excel with AI
Most of the small to medium sized companies are moving away from disparate excel sheets between teams to using advanced analytics, AI and Big Data solutions. As the cost of computing and data storage has tremendously come down over the last several years, even small to medium sized enterprise solutions are able to use AI based forecasting
360 degree demand sensing
With data proliferation the ability of the supply chain to adapt to demand changes is more complicated than ever. Successful companies are adapting by building systems that can pro-actively sense demand from multiple channels including external macro economic factors like weather, consumer sentiment etc.
Automation / BOTS
With the advent of IoT devices and BOTs, demand is changing in real-time on the factory floor. Automated bots can pick and fulfill demand through automation. More and more manufacturing companies are adding capabilities to address real- time demand planning and forecasting to complement automation in fulfillment.
Why is demand forecasting crucial for business?
Improved strategy for optimal pricing
Understanding demand for products at any given time can help price it appropriately. For example, if demand forecasting predicts softer demand for product or goods, reducing the price could help clear the excess inventory and vice versa.
Budget & production planning
In manufacturing it’s hard to prepare a budget without demand forecasting. Without forecasting production planning is not possible. When there’s an opportunity to invest in a new product line, or expand existing, accurate demand planning is key.
Improved customer satisfaction
If proper inventory levels are not maintained customer orders will not be fulfilled in time leading to customer satisfaction issues. Customers might look to competitors and businesses can loose customer permanently.
Reduce excess inventory
Excess inventory increases the storage costs for the inventory. Also, the longer inventory is in storage, the more likely it is to decrease in value. Demand forecasting can help reduce cost of inventory purchase and storage costs by having the right inventory at the right time.
Avoid under-stocking and expedited shipping costs
The flip side of excess inventory is under stocking. If the inventory is low and customer demand is high businesses tend to expedite shipment to meet customer expectations increasing the costs to business.
Demand forecasting methods
Factors influencing demand forecasting include time horizon and type of forecasting. Firms should ensure they consider these factors prior to embarking on a specific solution. A brief description of these categories follows.
Forecasting method consideration
Data collection
Depending on availability of data types, firms should consider applying proper forecasting methods. Ideally both qualitative and quantitative methods should be considered for a more reliable and robust forecasting.
In our upcoming blogs, we will discuss the solutions we offer and how effectively a retail business can use ML tools to predict the demand in a volatile market.
