A significant portion of retailers’ annual profits are attributed to the festive shopping season – from Black Friday, to the weekend before Christmas Day – which generally spans over approximately three weeks.
This year, however, an extra week of festive season spending means that retailers need to reassess their consumer demand prediction methods or risk being caught unprepared, understaffed and out of stock when shoppers pile in.
This is according to Vian Chinner, a South African innovator, data scientist and CEO of Xineoh – the Canadian machine learning company helping businesses to out-predict their competition, maximize their efficiency and enhance their customer satisfaction.
Chinner explains that the 2018 shopping season will be completely different to any experienced by retailers before.
“With 1 November falling on a Thursday, Black Friday will be taking place before pay-day for most people. The last weekend before Christmas, which marks the official ‘end’ of the shopping season, will then drag the season out over a period of four weeks and two days.”
He adds that, with the extended shopping season transforming consumer shopping behaviour over this period, traditional consumer demand prediction methods used by South African retailers will be thrown off completely and that more modern, artificial intelligence (AI) technology should be considered to ensure accurate predictions.
Chinner says, “With our experience in the e-commerce sector, we know that the structure of this year’s festive shopping season will see consumers spending more in the run-up to Christmas rather than spending the same amount, spread out over the longer period. This presents retailers with an opportunity to boost sales – and profits – even further than they normally would at this time of year, but only if they are prepared.”
However, he explains that most South African retailers still use prediction models which are based on calendar dates like the Autoregressive Integrated Moving Average (ARIMA) model which was developed by statisticians, George Box and Gwilym Jenkins, in the 1970s. “Basing predictions on dates would normally be relatively adequate, but not when we look at an outlier season like the one approaching.”
“For example, the 23rd of November is normally a really poor day for retail sales. This year, however, with Black Friday falling on this date we will probably see about four times the average number of sales for this day. Those using calendar-based prediction methods may not have the staff or inventory in place to cope with, and optimise, demand,” Chinner adds.
So, how can South African retailers accurately predict demand when the festive season is just around the corner?
Chinner highlights that a modern AI solution, like Xineoh, provides accurate predictions, quickly (within just two weeks of receiving a retailer’s data) and at an affordable cost.
He says, “Going far beyond estimating demand relative to calendar dates, Xineoh’s algorithm considers all variables specific to the current situation.”
“Matching people with products, inventory with opportunity, price with spending propensity and people with usage patterns, adopting a modern AI solution ahead of the surge in year-end shopping can help retailers make informed decisions and provide their customers with the service, products and experience they expect – no matter how the shopping season is structured,” Chinner concludes.