Gaining a competitive edge in seasonal retail

The profits of many firms, ranging from large retailers and supermarkets to small shops and cafés, depend greatly on not only the season but also their ability to make predictions about peaks in demand during specific weather conditions. The weather can be so unpredictable in the UK that the accuracy of such predictions can have a huge impact on profit margins. Research carried out by Tesco concluded that an increase in temperature from 20°C to 24°C leads to, on average, a huge 42% increase in the demand for hamburgers. Meanwhile, a major UK ice cream producer has reported a 1.2% increase in demand for ice cream for every 1% rise in temperature. So why has utilising this information not been capitalised on more?

Currently, the weather forecast is often underused by firms when making predictions and when used it is only in a basic way. Research published by the Met Office last October concluded that, out of the 200 large businesses who were surveyed, only 39% use the weather forecast (of which only 27% use commercial forecasts) in order to make predictions. This is despite the fact that 47% of the survey participants claimed that the weather is the third most important factor that drives consumer demand. In fact, the report showed that weather forecasts rank third bottom for factors incorporated into demand prediction, yet it is third top for factors which companies would like to use – especially as 67% of the participants claimed that forecasting demand is getting more difficult.

Many retailers rely only on a small set of weather variables based on the commonly available weather forecasts, generally relating to temperature and the level of precipitation. Yet there are many others which could also be used such as wind speed or humidity. It is likely that incorporating such data into the prediction for peaks in seasonal demand would allow for much more accurate predictions. Research has suggested that there is the potential to develop statistical models, outlining the complexities of how weather variables interact (e.g. high humidity accompanied by high temperatures can be detrimental) to impact retail demand, rather than the previous simplified approach that if the temperature exceeds a certain value then there is bound to be a demand for BBQ products. Moreover, there are often regional differences on how consumers respond to changes in the weather, and according to The Weather Company “barbecue sales triple in Scotland when temperatures rise above 20°C. In London, however, the figure is 24°C.”

So have any companies caught on to this? One company who has tried to create such a model is Tesco. Tesco have developed complex demand prediction systems based on 15 years of research into patterns of consumer behaviour under different conditions. Weather forecasts are taken for 15 different locations throughout the UK and as a result, orders from local distribution centres are adjusted automatically during the day.

Such a complex system as Tesco’s is unlikely to be financially viable for smaller businesses, due to both the expense of acquiring the appropriate data and the expertise needed to analyse it. This is where the developing areas of big data and data science are beginning to play a role in the form of the development of a new range of services by specialist companies. Such companies are now working with retailers of all sizes to help them better understand their markets and enable them to react effectively to changes in the factors - such as weather - which influences demand for their products. With access to the right tools and information, retailers potentially no longer have to rely on crude methods to guess their marketing strategy or level of production. Instead their strategies would be informed from a whole new level of insight which can only be seen by modelling complex interactions of the weather and consumers’ demand on a large scale.

Whilst the weather conditions may have minimal impact on the overall levels of demand in the economy the ability of firms to make accurate predictions about the levels of demand for specific products is essential for maximising economic welfare and minimising the waste of scarce resources. The simplest example being that increased demand for iced coffee results in a fall in demand for hot coffee. Ultimately, as demand prediction is becoming more and more significant in maximising firms’ profits, it is only those who can successfully capture the required data, and act accordingly, who will survive in the long run – the use of only simplified approaches (e.g. only using temperature) is not going to be sustainable in an increasingly competitive world.

Picture source: