Christmas Sales Prediction
We are immersed in the Christmas holidays. In Spain, traditionally, the season kicked off with the December long weekend, when everyone takes the opportunity to put up the tree, decorate the house, and take out their wool sweaters, and institutions proceed with the eagerly awaited ‘lightings’ of towns and cities. However, in recent years, the date has been brought forward with the consolidation of Black Friday in Europe.
Some time ago, this North American tradition timidly broke into the old continent, eventually consolidating itself as one of the most anticipated sales campaigns of the year by shoppers. In its favor is the date on which it takes place, the last Friday of November, already close to the Christmas period, which necessitates family spending on gifts and clothes to wear during the holidays.
Thanksgiving night used to be the starting gun for the most consumerist time of the year in the United States, and now it is in most parts of the world, although, like everything in 2020, the development of the campaign has been different, as will be Christmas sales. Lockdowns and recommendations to avoid enclosed spaces and contact with other people have turned Black Friday into a Cyber Friday, with massive online purchases, which are now extending to the acquisitions for our most traditional holidays.
In this atypical market, marked by a severe economic crisis derived from the health crisis, and by changing consumer habits, how do companies face this productive and logistical challenge? The best way is to partner with new technologies. The combination of Artificial Intelligence, Algorithms, and Advanced Data Analytics allows for the correlation and analysis of multiple variables, both internal and external, to determine the estimated demand for items and services. In volatile environments like the one we are in, predictive models show, with high levels of reliability, what is to come, facilitating decision-making throughout the entire supply chain.
Will sales fall? Which shopping channels will be most used? What products will consumers request? How much stock do I need? These questions have been latent in all companies since the beginning of the pandemic; they were already there before, as uncertainty is the faithful ally of the word ‘market’, but now they are continuous as the daily feeling is one of treading on quicksand. Complex situations serve to highlight, even more, that intuition, experience, and the results of previous campaigns (the model used until now) are no longer valid as a reference. Evidence is needed in the increasingly complex business decision-making process.
Converting Data into Information
The certainties we speak of are provided by information, and this arises from data exploitation, from objective inputs that, when interrelated, offer a future radiography of consumption trends. Data is a source of wealth that, until recently, was discarded without being fully exploited, and often without even knowing it existed. Every second, all processes in the supply chain generate information, so the first step for implementing demand prediction systems is to compile this data.
Once the raw material from Internet of Things (IoT) devices, ERPs, MRP, WMS, CRM, etc., has been collected, it is time to interrelate it. Through advanced predictive algorithms, data engineers connect the extracted series to convert them into useful and objective information that is continuously updated. For these results to be easy to interpret, the next step involves another disruptive technology, Business Intelligence, which allows information to be displayed in visual and intuitive dashboards. With these, personnel, according to their areas and permissions, can efficiently guide business actions, moving from working with estimations to working with predictions.
From Estimation to Prediction and from Prediction to Prescription
But there is still one more step: the leap to prescription. Machine Learning allows for the definition of predictive analysis models, based on the evolutionary data of product consumption that have a sales history, their substitutes, or those with similar characteristics. In other cases, behavioral estimation is performed based on associative models, grounded in market trends, purchasing power, or the influence of marketing campaigns.
Based on this information, prescriptive models self-adjust, through machine learning processes, for deviations caused by atypical periods like the current one or by changes in consumption trends. The great advantage of these sales prediction systems is, consequently, the generation of models that anticipate the reaction of the potential market, identifying future consumption needs, with which to prescribe demand planning.
Having forecasts of the estimated sales evolution allows for maintaining necessary stock levels, avoiding stockouts (potential loss of customers) and overstocking (capital immobilization); it enables optimizing transport routes; reducing risks such as obsolescence of references; lowering operational costs; conducting marketing and pricing campaigns to reinforce low-rotation periods; improving agreements with suppliers thanks to higher levels of foresight; granting reaction capacity to demand variations; and saving planning time that can be dedicated to decision-making, which will also be much more accurate, among many other benefits.
Efficiency, Security, and Profitability
Ultimately, demand prediction translates into efficiency, security, and profitability, and that, in any market, but even more so in current circumstances, is golden. The Covid-19 crisis has accelerated the need to embrace digital transformation if one wants to remain competitive. Those who have started the path of digitalization and have data-driven strategies possess a solid foundation for making sound decisions, eliminating high levels of uncertainty. Atypical Christmas campaigns like the one we are experiencing will be better faced by those who, for example, have the demand prediction systems we have discussed, which transform questions into certainties, allowing for efficient supply chain management.