How can retailers best use Big Data?
Big Data has numerous retail applications, and in the August issue of National Liquor News, Norrelle Goldring deciphers them.
“Information is the oil of the 21st Century, and analytics is the combustion engine.” – Peter Sondergaard, Gartner.
According to Wikipedia, the term ‘Big Data’ has been in use since the 1990s, and 2002 is considered the beginning of the ‘digital age’ with the move to digital storage and away from analogue storage on devices such as tapes.
But what is Big Data, and how can it best be used in a retail context?
What is Big Data?
Big Data typically refers to large datasets computationally analysed to reveal patterns, trends and associations, including relating to human behaviour and interactions. Datasets are increasing rapidly with the advent of mobile devices, smart devices (The Internet of Things [IoT]), and other data capturing mechanisms such as sensors, cameras, microphones, beacons and wireless sensor networks.
Originally associated with the three concepts of volume (large amounts), variety (format variety including numeric, text, images, video), and velocity (frequently updated), current definitions of Big Data refer more to the use of predictive analytics, user behaviour analytics and other advanced analytics methods.
A 2011 McKinsey Global Institute report characterises the main components and ecosystem of big data as being techniques for analysing data (A/B testing, machine learning, natural language processing); technologies such as cloud computing and databases; and visualisation such as charts and graphs.
How can Big Data be used in a retail context?
Big Data has ‘seamless’ and omni-channel applications. Google research suggests that 98 per cent of Americans switch between devices in the same day. Retail data gives retailers and brands the ability to extract insights across devices and touchpoints in order to create seamless campaigns and offers across multiple channels including in-store, web, live chat, email and mobile. A study by Aspect as far back as 2014 suggests that retailers who adopt omni-channel strategies have 91 per cent greater year-on-year customer retention rates.
Some of the uses of big data in retail include:
Demand: by understanding customer habits, retailers can understand which of their products and services are most in-demand in real time (top sellers are also an indication of broader trends) and which ones they should potentially stop offering. It enables monitoring and prevention of out of stocks, and smarter purchasing and ordering from suppliers including setting automatic quantities, as well as seasonal items and cross-selling opportunities and time of day/week offers.
Prediction: trends to inform popular products, and particularly spend which is collected through loyalty programs and credit card transactions as well as IP addresses and user logins. Analysis of spend patterns enables predicting of future patterns as well as the ability to customise offers, and the identification of which customers are most likely to be interested in which offers and thus likely uptake. Other data such as the weather, concerts, political events etc. can be analysed to predict demand for liquor types.
Pricing: which prices yield the best results on particular products, and discounts optimisation by SKU.
Customer journey analytics: by understanding touch points, customer device preferences, and online browsing behaviour including social media, retailers can understand where customers look for product information, where customers are being gained and lost, and the most effective channels to reach them.
Personalisation: create individualised customer recommendations and offers based on their purchase history. Tailor merchandising, deals, and marketing campaigns. As the IoT increases, more retailers may equip stores with sensors which enable offers and new product information to be proactively sent to local passers-by with the app installed on their mobile device.
Market analysis: such as competitor catalogues to identify newly launched products, and the most in-demand or promoted products.
Customer sentiment: via text mining and other algorithms to understand level, performance and drivers of customer satisfaction.
Customer profiling: identifying customer lifetime value, which customers have the highest value and which other types of customers are most likely to replicate this profile. Creation of individual customer personas in order to tailor offers and communications.
Operations: analysis of data to optimise supply chain functions including shipping, staff, pay, theft, breakages etc.
How it looks in the real world
Starbucks: can predict the growth potential of every individual new store by looking at metrics such as location, traffic, area demographics and customer behaviour. Insights from their 90 million transactions per week was used to develop innovations such as a tailored digital rewards scheme that becomes more intuitive the more data it gathers.
The Weather Channel: partnered with haircare brand Pantene and US drugstore chain Walgreens in order to anticipate when air humidity would be at its highest, enabling a targeted campaign to prompt women to prevent hair frizz. This resulted in a 10 per cent increase in Pantene sales at Walgreens for two months around the campaign, and Walgreens saw a four per cent sales lift across the total hair care category during that same period.
Amazon: possibly the originator of recommending items for customers based on past searches and purchases, they have generated 29 per cent of sales through the recommendations engine, which analyses more than 150 million accounts.
Costco: when a Californian fruit packing company warned Costco about the possibility of listeria contamination in fruits like peaches and plums, Costco emailed specific customers who had purchased the affected products, instead of a blanket email to the entire customer base.
Kroger: uses analytics to determine which products an individual customer actually wants to buy, then sends them customised digital coupons for those products.
And in liquor:
Heineken: has been using Big Data for a number of years in a variety of ways, ranging from knowing where in a Walmart store a six-pack was picked up, to smart social media campaigns and a smart beer bottle that dances to the rhythm of music at parties.
On-premise: Israeli start-up Weissbeerger has developed analysis via sensors in brewing equipment that are transmitted to a computer via Wi-Fi, and then visualised on an app used by bar owners. Algorithms analyse which beer should be on a discount and at what period, resulting in an increased profit of up to 80 per cent. Insights include which brands are more popular on what dates and what times of day specific beers are most consumed, allowing for smarter ordering and promotions.
Considerations and challenges
Big Data isn’t all ‘beer and skittles’, however. One question for companies embarking on the Big Data journey is determining who should own Big Data initiatives, particularly those impacting the whole organisation.
The other is obviously private and data breaches, of which there are numerous recent examples. One of the more famous privacy examples was the Target USA example in 2012 of targeting pregnant mothers before they share their baby news, which resulted in a father finding out his teenage daughter was pregnant before she had told him.
Get yourself a CRM system, if you don’t already have one. Start collecting customer data. Get a retail software analytics platform or provider to help you if necessary. If you have the resource, hire a data analytics and/or research person or team to mine and interpret it and tell you what you should be doing, to whom, where.
Most importantly, begin with the end in mind. Rather than trying to boil the proverbial ocean, determine what areas you most want to optimise or improve and work on those. The most effective big data strategies identify business requirements and objectives first, and then tailor the infrastructure, data sources, analytics, resources and suppliers to support the objective.