Data Science in Retail: Three ways analytics can help you plan

Media release

Rich Pugh, Chief Data Scientist at Mango Solutions looks at utilising data science in retail to help plan ahead and get through these challenging times.

11 June 2020 Posted in  A1 Retail

Retailers who have been shuttered since March are having to consider making significant changes on a number of levels when many re-open. Even for businesses that, on the surface of it, seem to have thrived (think grocery and e-commerce businesses) actual profits are likely to have remained flatter than many might think, due to increased costs in the supply chain, logistics, resourcing, and investment in additional worker protection and sick pay. Simultaneously, profiles of consumers themselves may be changing. All of this creates a backdrop of uncertainty, where retailers have to adapt to a “new normal” that we cannot yet fully understand.

Data Science in Retail: Planning for “what if?”

Right now there are several open questions that matter to retailers – including, but not limited to – “will working from home become the new norm – and will this change how and when people shop?”, “when can stores fully reopen?”, “will customers still shop the way they did before?”, “what will the new competitive landscape look like”. With a recession looming on the horizon, retailers need to make effective decisions to deliver budget and resources to the right initiatives. The question is, which initiatives are the right ones?

The truth is, there is no absolute right answer here; there simply isn’t enough clarity on what the future holds. Retailers need to use the data they have on current events, historic trends and consumer behaviour (mixed with knowledge and insight about the current environment) to build models to explore “what if” scenarios. Building these kinds of models, particularly at short notice and for smaller outlets, takes time and considerable data engineering skill to bring together the different data streams into a cohesive solution that can visualise the best outcomes for different situations.

Successful models that can demonstrate what success would look like in different scenarios can be an invaluable tool, providing retailers with a way to prioritise investment across common trends. In addition, such models can help with resource allocations. The right models, used correctly, can help inform strategy on where shops should start reopening, at what scale, with what conditions in place, and with what staff and skill sets. All of this will be essential for effectively re-starting retail for the “new normal.”

Understanding your “new” customer

The second consideration is that customer behaviour and concerns have likely shifted – and there’s no way to guarantee it will revert back to normal once the lockdowns are eased. Rising unemployment and talk of a recession may have customers on edge about making impulsive purchases, for example, which means changing the way you sell, and market, to address this. But step one is understanding who your new customers are, what their pain-points are and what their likely behaviour will be.

While the process of mapping personas is not a new one, the current situation throws a spanner into the works and relying on single data streams is now likely providing an incomplete view of the situation. For example, existing customer persona mappings might have worked on making product suggestions based on a user’s purchase history, but now their willingness to try out new products may have decreased. This means trying to bring in everything from market research data to transaction history to social media trends to map sentiments and trends and understand how to interact better with consumers in the brave new world they find themselves in. Investing in data science approaches to bring together these disparate data sources, and the right analytics support to understand them, will be critical to selling to this new wave of mid- and post-Covid shoppers.

Customer acquisition

With an understanding of who your new customers are, what concerns your existing customers might have, and the right models to build strategies for different scenarios, you are now in a position to do what successful retailers do best – convince customers to buy their products.

The question now becomes how you do this most efficiently. Modelling the best approach to customer (re)acquisition can be challenging given the lack of historical data on the current situation. However, combining this trend information with ongoing testing can help retailers know how, when and where (online or in store, for example) to engage with customers to convince customers to buy, increase order value and promote repeat business and customer re-engagement. By A/B testing different messages and strategies, as well as bringing in historic data and external behavioural analysis to elucidate any trends appearing in the data, retailers stand to benefit from more, happier, customers. This can also reinforce and inform future modelling as to what business strategies are and are not working in the current climate.

The COVID-19 crisis is undoubtedly devastating for many – both on personal and professional levels – but it also presents the retail sector with an opportunity, an imperative even, for innovation. This task may seem daunting but there are resources that can help. Working with the right provider can help retailers bring data science to the heart of their business faster, not only navigating the challenges of today, but preparing for a data-driven future with the right skills and resources throughout the organisation to plan for any “what if” scenario.

Written by Rich Pugh co-founder and chief data scientist at Mango Solutions