In the last few years, a lot of retailers are experiencing significant variation in demand as Supply Shortages and price increase drive more erratic customer buying behaviour. In these market conditions, a lot of retailers are focusing less on perfecting their long-term forecast and focusing more on responding to the error in their forecast. One tool that is helping retailers with this focus on agility is short-term demand forecasting. Short-Term Forecasting, otherwise known as Demand Sensing, is a popular approach that leverages near-real time data streams and machine learning to provide more accurate short-term forecasts. Short-term forecasts are especially useful for retailers who make critical operational decisions, such as allocation or distribution, within 3-10 weeks of the demand.
What is Short-Term Demand Forecasting?
Demand Forecasting is a type of data analysis that is used to estimate future demand for products and services. As broad as that sounds, it means that Demand Forecasting is a technique that can be quite flexible depending on what you want to use it to achieve, the kinds of data you have on hand, and the kinds of demand that your products have (and you can find out all about that here). And, this is where short-term Demand Forecasting comes in.
Since Demand Forecasting can vary depending on how far into the future you want to forecast for at a given time, short-term Demand Forecasting is generally used to forecast demand for periods anywhere from 0-3 months at a time. This can be stratified even further by the purpose of the forecast and operational cadence of your business.
Grouped in with short-term forecasting is Demand Sensing, which is specifically used for forecasting on the hourly or daily scale up to 10 weeks into the future.
Why do you need it?
As we mentioned earlier, short-term Demand Forecasting can be used to achieve a host of different things in the short-term - think everything from preparing sales-boosting policies to planning your product distribution. And, these things can be split into two major groups:
- Tactical forecasting - think inventory and short-term capacity planning
- Operational forecasting - includes planning transportation and production, and inventory deployment
Generally speaking, tactical forecasting is done at the monthly or weekly level, whereas operational forecasts are produced for anywhere from an hour to a few days at a time. In addition to tactical forecasting, monthly and weekly forecasts can also be useful for promotion planning.
Additionally, short-term forecasting can enable you to incorporate factors that you wouldn’t be able to otherwise with long-term forecasting. Variables such as foot traffic, weather data, and other sources where the data is at its most accurate point in the short-term can be included. Plus, it can be especially useful for forecasting demand in the midst or wake of significant events - think everything from financial crises to natural disasters and pandemics.
For instance, let’s consider a retailer selling computer monitors, stationery, and other office equipment during COVID-19. To understand how your customers have been affected by the pandemic (and what that will do to your demand), you can analyze POS sales data, customer orders, web traffic on your website, or any other metrics that are available and capture changes in customer behaviour. And, you can combine this with COVID-related data - think virus trajectory, lockdown periods, and other macroeconomic indicators - to give you an insight into what might happen in the short-term. From this, you might notice that the demand for webcams and headphones start to spike at the start of lockdown periods (and the onset of video conferencing and other joys of working from home). This insight can be useful for deciding how much of a product to send to which location. Whether it’s COVID-19, a financial crisis, or any other large-scale event, injecting this information into your short-term forecasting can increase the predictive power of your models, especially where these events cause major disruptions and make all of your other sales history irrelevant. And forecasting in the short-term during these events can give you the agility to make quick decisions as conditions change.
What You’ll Need & Challenges to Expect
When it comes to creating useful and accurate short-term forecasts, it can be a hard and dangerous road. Lucky for you, there are some tools and tricks that we’ve picked up in our travels and they’re available to you at a price (but, we only accept time spent reading this article and witty comments as payment).
The key thing that you’ll need to create short-term forecasts (or any kind of forecast, really) is data. The bulk of your data is probably going to be historical sales data, but you can also include:
- Web analytics
- Environment data
- Historical shipments and future orders
- Competitor pricing
- Mobility data
Basically, you should include data that can be used to identify trends and patterns in the demand for your products. For instance, you’ll need to consider seasonality, promotional effects, and even causal factors such as price or the weather when creating your forecasts, and either incorporate them into your forecast (if you’re using statistical methods) or use software (such as Remi’s Demand Forecasting platform) to worry about that for you.
But, the reliance on data also comes with challenges. Specifically, you’ll need to ensure that the data you use is clean, reliable, and relevant to the demand forecasts you’re producing. This minimises the errors that can arise from dealing with dirty data as well as reducing the noisiness of your data - making your forecast easier to interpret and actually use.
In addition, it’s important to remember that your data will become less useful if you use it beyond the time that it is accurate for. For example, using a weather forecast for a given week can be useful for forecasting around the period that is forecasted, but it becomes less relevant if you continue to use it once that week has actually happened. To avoid this issue, you should ensure that your data is updated regularly.
And, as exciting and useful as short-term forecasting is, focusing all your energy on the short-term and neglecting your mid- and long-term plans can risk breakdowns in other areas of the supply chain or lead to long-term goals not being met.
As long as the world around us is prone to change, Demand Forecasting will be your best bet for understanding and predicting what your customers want and what they will do. By integrating short-term forecasting into your operations, you can make informed decisions to meet customer expectations while generating a profit.
Want to find out more about Demand Forecasting? Why not have a read through of our case studies, where you’ll find out how we’ve used demand forecasting to help increase stock availability and improve the accuracy of forecasts. Or, check out our blog for the latest AI reads here. Once you’ve sated your hunger for knowledge and you’re ready to take your Demand Forecasting strategy to the next level, drop us a line here.