Covid-19 has transformed customer behaviour from the way people shop to the products and services they buy.
These changed shopping patterns are likely to last into the near future. They may even extend into the longer term because, on average, a new behaviour persisting for about two months becomes a habit.
In order to build and retain a sustainable business, retailers urgently need to determine the extent to which these changes in consumer behaviour will affect their product categories and brands, as well as how they deliver them.
The current economic environment has made it harder for retailers to recover from bad decisions based on inaccurate forecasting. So the big question is — how can retailers improve demand forecasting?
Throughout Covid-19 restrictions, businesses and individuals increased their social media usage. Social media data offers an opportunity for all businesses to improve their demand forecasts. Social media analytics provide insights that retailers can leverage.
Social media can also be an effective sensor when it comes to receiving signals from current as well as potential customers.
Results from a study carried out by researchers from Lero and Maynooth University, and published in the, show a correlation between customers’ opinions on Facebook and Twitter, and actual sales.
At the onset of the Covid-19 lockdown, we saw in most countries an unprecedented level of panic buying for toilet rolls and other everyday essentials, and many retailers were seen struggling to mobilise their supply chains in time to fill the empty shelves. We saw a similar trend with hand sanitiser and face masks.
One of the main reasons for the empty shelves was that most retailers are stocked based on demand estimates from previous weeks and years. Social media analytics could have helped in this scenario with forecasting demand based on analysing the customers’ sentiments.
Informal style, casual language, and the use of special symbols in social media posts make it challenging to extract actionable data. The noise in social media data hinders accurate predictions about overall demand.
However, with advancements in social media analytics research and the level of granularity possible in terms of predicting demand for each product separately, information which is collected can be converted as a demand-forecasting and market or trend-sensing tool.
The major factor in extracting value from social media is to apply multiple data-cleaning techniques in conjunction with one another, so the data used in the forecasting models represents the actual state of affairs.
For example, analytics techniques such as sentiment analysis, when applied in conjunction with topic modelling, can turn the random posts, comments, and reviews by customers on the social media platforms of retailers into actionable data.
It can help in quantifying the customers’ sentiment for product categories and brands to help retailers to improve forecasts and increase sales.
However, valuable social media data is no substitute for historical sales data.
For example, retailers can track a social campaign through the launch of a product and forecast demand by combining the historical data with social media data.
Retailers can either choose to build or improve in-house demand forecasting systems by adding social media analytics to the already existing forecasting system or can use third-party tools.
Examples of third-party tools which use social media data for demand forecasting are Trendscope and Simporter. Small Irish retailers can also engage with research institutes using Enterprise Ireland Innovation vouchers to start building capabilities for social media analytics.
Covid-19 has created many obstacles and challenges for retailers, but it has also opened some windows of opportunity. The survival of retailers will most likely depend on how they leverage these opportunities.
Social media data is one of these opportunities, and it is imperative for retailers to take advantage to remain relevant in the “new normal” world.