A Deluge of Data Behind Black Friday
  • 4th December, 2018
  •  

A Deluge of Data Behind Black Friday

By Megan Yates

Ever wondered how companies sell products at such low prices during annual sales events like Black Friday and Cyber Monday? Well big retailers don’t set these deals by chance. They rely on the power of data to inform everything from which products are discounted, and by how much, to how deals are communicated to customers.

In 2017, 4.7 million card transactions were processed in South Africa on Black Friday and Cyber Monday alone, amounting to R2.5 billion ($182 million) worth of debit and credit card retail transactions. That was up from R2.3 billion ($168 million) in 2016 and was twice the daily average, according to BankservAfrica. In the US, Black Friday sales were around $5.03 billion in 2017. This year Alibaba set a new Singles Day record with over $30.8 billion in sales in the first 24 hours. In a competitive retail environment, it’s imperative that companies take advantage of the hype surrounding these annual events, and do so intelligently.

Big retailers like Alibaba and Amazon gather huge amounts of data on customers throughout the year. As consumers view items, purchase, add and remove from their carts, so this behaviour is recorded and stored. Even customers that purchase from a physical store often view and research items online ahead of visiting a store and increasingly customers use their mobile phones in store to check prices and make buying decisions. This is all invaluable data for retailers.

When planning for major events like Black Friday and Singles Day, retailers must assess available inventory, historical product performance, current purchasing trends, historical event trends, economic factors and even the weather. Retailers crunch their internal data and add in external factors to predict sales and determine what volume would still result in a desired ROI to ultimately inform discounts. Retailers need to determine:

  1. Which products are likely to be in greatest demand?
  2. Which types of offers are most likely to generate the most interest?
  3. Which types of communication channel and messaging will achieve the best response?

With machine learning techniques we can take things a step further and predict the amount a specific customer is likely to spend on various products. Modelling at a customer level like this allows a further layer of intelligence when determining deals.

Ixio recently created a pricing model to forecast sales trends under various pricing scenarios for a large retailer. The model took into account historical prices, historical trends, inflation, consumer price index and other economic factors. The client implemented certain price changes and we could track how our forecasted volumes performed against observed volumes. Our forecasts across multiple products were within 1% of observed values. With confidence in model performance, the client can now make price adjustments with more knowledge of how this is likely to affect volumes, and can also discount accordingly. 

Retailers armed with more data and rigorous, science-based approaches stand to do well in a highly competitive environment.

(Photo by Anna Dziubinska on Unsplash)

Share This Artcle :

About Author

Megan founded Ixio in 2012 after seeing a need for strong, data-led modelling and analytics in business. As Chief Scientist at Ixio Analytics, Megan leads our advanced modelling programs and coordinates the technical requirements for our clients. Her background in Evolutionary Biology allows her to bring a rigorous scientific approach to analytical problem solving. Megan is highly regarded in the industry and contributes regularly to data and analytics conferences, both locally and internationally. She takes a keen interest in environmental issues and is an avid surfer.

Comments