Sales Forecasting With Weather Data
Sales Forecasting with Weather Data. Art, Science or a Fools Errand!
The weather affects 70% of companies globally, with weather variability costing more than €560bn in Europe. And an estimated US$2 trillion worldwide for businesses operating in retail, consumer goods, apparel, transportation, utilities and food processing.
Climate change will make the winters in the UK wetter and the summers more prolonged and extreme. According to weather ads, in the USA 33% of business activity is sensitive to weather fluctuations.
A fall or increase in seasonal weather temperatures by just 1% can have a significant impact on retail sales. In the UK adverse or volatile weather is estimated to cost the retail sector approximately £3bn.
To illustrate in the winter of 2014-2015, the unusually warm temperatures, across Europe and the USA, reduced the level of consumer spending. Leading to lower apparel sales, a delayed launch of the spring season for H&M and store closures and job cuts.
Failure to incorporate variations in weather into the analysis of retail sales may lead to biased demand forecasting and a misinterpretation of financial indicators. The effects of weather on apparel and sporting goods retail sales was demonstrated to be large and significant in a 2016 paper by Brigitte Roth Tran.
Also, the most extreme forms of adverse weather shocks reduced sales by an average of 22% and 12% at stores and outdoor and indoor malls. The most favourable forms of weather increased sales by an average of 16% and 10% in outdoor and indoor malls.
Furthermore, when regarding ecommerce sales, as temperatures decreased the number of consumers staying at home to shop increased. However, the volume of online sales is subject to geo-specific consumer purchase behaviour. Hence, sales volumes are influenced by weather patterns that are not usually experienced within a geographical location.
In other words, according to Japanese e-commerce company Rakuten, online sales volumes are closely linked to weather patterns that are not usually experienced, and the attitudes and culture of the local populations.
The impact of weather variability, on the US economy was reported in a study by J.K Lazo, to be as much as US$ 485 billion in 2008. Or 3.4% of the USA’s total domestic product.
Dutton in 2002, stated that one-third of private industry activities representing revenues of US$ 3 trillion, were exposed to weather and climate risk.
How to Overcome Weather Risk
The accuracy of weather forecasting and the incorporation of weather data into determining consumer demand. May become an essential part of the toolkit of the retail sector and many other industry sectors. In a study by J.K Lazo in which eleven sectors in the USA were analysed for the impact of the variability in weather on the sectors.
The retail industry was one of the sectors with the lowest levels of weather sensitivity when compared to for example the FIRE sectors. Finance, Insurance and Real Estate or the wholesale industry.
As the weather affects a broad array of industry sectors such as agriculture, manufacturing, financial services and construction. One method that firms have adopted to mitigate against weather risk is through deploying weather derivatives as a hedge against adverse or volatile weather conditions. Alternatively, companies can incorporate weather data into their sales forecasting.
There are a significant number of events with short lifecycles, inherent within major annual trends. Organisations able to exploit these narrow windows of opportunities profitably win. Retailers are beginning to realise this, and are adopting an empirical approach to incorporating weather data into sales forecasting.
To illustrate, Sears, noticed that car battery sales increased after three consecutive days of sub-zero temperatures, and thus advertises on the 4th day (Weather Unlocked 2014).
The release of smart menus by McDonalds or weather-related sales promotions by Campbells is also an illustration of retailers beginning to recognise the opportunity that leveraging 3rd party data in this instance weather data. To exploit windows of opportunities hidden within major annual trends.
Machine Learning and Deploying Weather Data for Accurate Sales Forecasting
Weather accounts for 3.4% of sales variations according to Lazo et al. (2011). Understanding how weather influences sales is essential for several reasons;
Managers need to account for actual weather when analysing past sales in an ex-post analysis. An ex-post analysis is reviewing financial sales from the past to predict likely future sales.
To understand the different factors that impact historical sales. Thus, as sales of clothes are significantly affected by good or bad weather, revenues and sales quantities should be adjusted to enable an un-weather-biased planning for the future. To improve order management and avoid unnecessary mark-downs.
Managers may also consider weather forecasts in an ex-ante analysis to anticipate the impact of weather on future sales and thus improve daily decision-making
The impact of weather predominantly in changing shopping behaviour is that it keeps customers from stores, diverts them to engage in online shopping, and influences the perception of fashion goods.
Steps for Creating a Model Using Weather Data to Improve Sales Forecasting
Data is sourced from historical weather data to develop an empirical Model.
Create a model to quantify the explanatory power of weather information on daily sales
Identify store-specific effects and analyse the influence of specific sales themes across a broad range of different product categories
Leverage insights to conduct ex-ante analysis using weather forecasts.
Analyse how the model can be used to improve the accuracy of the sales forecasts
Factors Linking Weather with In-store Sales
Comfortableness: When weather does not prevent customers from going to the store but creates a feeling that shopping will be unpleasant. Bad weather eliminates distractions and allows individuals to focus on their work instead of leisure activities. This phenomenon effects mood, automobile plant productivity and stock returns.
Physical Prevention: Weather induced hinderances such as snow drifts, which impede the ability of consumers to visit a store.
Psychological Effects on Shopping Habits: Bulk buying to hoard supplies as with Covid-19 pandemic or lower prices during extreme weather may trigger stock ups.
Desirability of certain products during certain weather periods. For example, umbrellas during rainy periods. Or barbecue grills during sunny and warm weather. This phenomenon is observed in the fashion industry.
Thus, cold winters advances the sales of warm winter clothes. Whilst sunscreen is sought during warm and sunny weather.
The 4 Theories of Weather
Theory One
Cold temperatures and precipitation deter consumers from travelling to stores. Shifts in consumer behaviour as a result of the weather is termed the “Convenience Effect” which states that “excessive cold and heavy rain will reduce sales of products that are easily deferred such as furniture and apparel” Convenience Effect except in a lockdown. Also shopping centre attendance is negatively impacted.
Theory Two
Weather may physically prevent consumers from shopping such a heavy snowfall. Which of course will direct consumers to shop online. Thus, poor weather will have an adverse effect on footfall to supermarkets or hyperstores located at the edge of towns. But will have a positive effect on online sales.
Theory Three
Psychological effect of weather on consumers causes changes in their shopping behaviour. Low humidity, high levels of sunlight, high pressure and high temperature are associated with good mood and that people in a positive mood are more likely to self-reward and spend money.
Theory Four
The weather effect varies across different product categories. The impact depends on the special characteristics of the product category. Hence hot weather advances the sales of ice creams, soft drinks and sandals, rainfall has a positive effect on umbrellas and raincoats.
The Magnitude of the Effects of Variable Weather on Retail Sales
There are 3 - theories on the magnitude of the impact of weather on retail sales
Purchase Timing Theory: Reduced sales in the current period will be offset by sales increases in future periods. Hence, the temporal effect of adverse weather only delays consumption to some point in the future. Sales for items such as seasonal garments, durable goods such as cars, consumer electronics and building materials are merely postponed and not permanently lost.
Permanent Impact Theory: This is the impact of sales on products bought impulsively. Thus, seasonality plays a role here. During hot weather ice cream and refreshing drinks are purchased impulsively. Hence if sales are not captured before the window of opportunity closes, the moment is lost.
Weather Consumption Cycle: Posits that the effect of the weather on the overall economy is more significant. Hence in countries like Morocco, absence of rain is associated with diminishing national output.
However, in the context of the retail sector theories one and two prevail. Hence for example, when Easter occurs early in the year overall sales are lower than when Easter occurs later in the year. As a result of variability in weather, retailers are exposed to two types of risk. The risk of overstocking and the risk of understocking.
One way to overcome this is to build and agile infrastructure and deploy data to leverage uncertainty as a competitive advantage.
