How does a retailer really drive value from big data analytics?
It is a broad theme, but one explored very effectively in the context of retail at this year’s NRF Annual Convention and Expo in New York.
A huge show attended by thousands of retail insiders eager for new revenue-gushing ideas was the perfect stage for a discussion of this crucial topic. After all, 75% of in-store purchases are influenced in some way by digital interactions, generating data that is the raw material from which gold can be extracted.
A major US retailer
Fortunately, at Teradata we have experience of putting this into practice and I was able to open a window to the future in a public discussion of analytics and big data in retail with Sameer Hassan, who is VP for e-commerce and marketing technology at Williams-Sonoma.
For those who don’t know, Williams-Sonoma is a highly-regarded US retailer that sells homeware and furnishings, much of it aspirational and upmarket.
We discussed a simple example of how analytics solved a problem which can impact retailers with a big range of sophisticated items on its website. Such a website will typically identify keywords from a customer’s search and serve back the results.
That’s fine if you are just looking for cereal on a supermarket website. Most of us know what cereal is and share a common understanding of what to call it and what it looks like.
However, Williams-Sonoma has a very extensive range of household items which makes searching less clear-cut.
Customers search for blankets, for instance, when the manufacturers designate them ‘throws’. Result? Not enough blankets were shown to customers and sales were missed.
Finding what they wanted
What happened was that Teradata set up a project to combine analysis of web-browsing terms with insights into where people actually went on the website, making it possible to ensure blankets and throws are both on display when a customer only types in one of the search terms.
Our joint success in this project drove up purchases because customers were better able to find something they wanted.
Similarly, Williams-Sonoma has used analytics to increase personalisation and extend what they offer each customer. This is not simply, for example, a case of targeting more baby-related items at customers whose searches indicate they have a new-born. It is about having the automated capability to match offers to changing requirements as the child grows into toddler and so on.
Knowledge is power
Collecting and analysing data in this way allows a business to build up tens of thousands of metrics about every customer, from their age to their purchasing history and web-browsing patterns. We don’t do it for fun. Such depth of knowledge makes predicting the effects of promotions and campaigns much easier, more accurate, more flexible and more effective.
Yet simply acquiring the insights from all this data, is not enough. Using this information to boost efficiency and augment revenue is the difficult part. It requires a business to scale up across all its business processes, perhaps feeding the information to a call centre, email marketing customisation algorithm or CRM system. This is another challenge in itself.
A cultural revolution
It requires the culture of the organisation to be one in which the marketing department is actively using the insights, rather than letting them stack up in a virtual corner. There are no excuses for such neglect, because the volumes of data involved in retail, though substantial, are perfectly easy to handle. Very often it is these cultural barriers that are the most difficult to overcome.
The leaders in this field are the pure-play online retailers, who were set up on data and have had to adapt to big volumes quickly.
More things equals more data
But in store retail, the volumes of useable data are fast evolving. The use of IoT technologies such as beacons and video, is generating more data from which operators can sweat value. Beacons, which track customers’ movements around a store by picking up their smartphones’ wi-fi signals, will indicate which areas and displays are popular at which times of day.
Video, on the other hand, has been used to monitor reactions to individual banners or pieces of signage, noting whether faces turn in the right direction.
All this data can be stored and subjected to analysis, particularly as the cost of storage is going down all the time.
Car parks – a new source
Indeed, one of the lessons retailers have to learn is that they must always be on the look-out for new data sources if they are to improve their operations and match customer requirements. Take the example of the supermarket operator in the US that realised it could intercept the feed from its car park security cameras.
It is relatively easy to automate the recognition of filled car spaces compared with black tarmac and white lines. The deployment of analytics on this data enables the store to be notified if there is a sudden rush in cars, perhaps because of the end of a sporting or school event.
From the analysis of its point-of-sale data, the superstore understood there was a 15-minute lag between the car park filling up and increased numbers of customers going through the tills. As a result of using analytics, the operator is now fully prepared for customer surges without having to cause unnecessary disruption. Staff are alerted in time, but not so early that operations such as re-stocking are badly affected.
It is an improvement in efficiency which makes life easier for customers, but in a way that is not necessarily obvious. Nonetheless, a competitor without this advantage runs the severe risk of gaining a reputation for being slow and poorly-managed.
Slow and poorly-managed – that is not how retailers can survive the many pressures they now face, nor how they can seize the huge opportunities to drive value from their wealth of data.