Data from loyalty cards, store transactions and social media is helping bricks and mortar retailers understand the who, what and when of sales, but when it comes to what’s happening inside the store, information is pretty thin on the ground. Until in-store cameras came along.
While traditional businesses still account for 95 percent of retail transactions, online sales are growing at four times their rate (according to 2012 U.S. Census data) due in no small way to the enormous amount of information they have on their customers.
Using a business strategy reminiscent of the corner store when merchandisers had personal relationships with customers, successful online retailers are using their data bank to greet shoppers by name when they enter the e-store, understand their shopping habits, remember their last few acquisitions, recommend purchases and dynamically customise the storefront to showcase relevant products. “Unlike traditional retailers, online stores automatically store information on customer behaviour,” notes Nils Rudi, INSEAD Professor of Technology and Operations Management, who is studying the impact of retail analytics. “They are able to look at not just what customers bought, but how much time they spent in certain sections of the store, what they looked at and what they didn’t buy.
“Amazon, for example, has spent enormous resources on observing and understanding customers which has led to their recommendation strategy. Now a large part of their volume comes from recommendations made.”
This type of cross selling and other category-penetration techniques using big data have the potential to increase sales by 20 percent and profits by 30 percent, as indicated in case studies by McKinsey & Company.
Keen to emulate e-retailers’ success, traditional retailers are now realising the value of such information. “All that’s needed,” says Rudi, “are the tools to improve data collection.”
Cameras track shoppers every move and interaction
Big retail chains such as Walmart, Sears and Target are investing heavily in computer analytic programmes like Hadoop, using information taken from store transactions, social networks, smart phones and Twitter feeds to track customers’ shopping habits. Rudi’s research aims to take this a step further using pattern-finding algorithms to analyse huge amounts of information taken from cameras, similar to security cameras, installed inside the building.
Working closely with a retailer in the Middle East, he has placed cameras in ten of their shops tracking customers and capturing their every detail: not just what they buy but how they are dressed, whom they are with, which direction they take around the store, where they stop, what they look at, touch, smell and hold, how they communicate with each other and how they seek help and interact with sales staff. Also of interest is what (and why) they don’t buy. Everything is efficiently noted and coded for analysis using an innovative system developed by Rudi.
The stores are located in diverse areas at different locations – from shopping malls to high street strips – and each has a different clientele, some targeting tourists, some locals, some expats, with each group showing distinct differences in shopping habits.
“The data is linked to the point of sales transactions, which lets us see what behaviour leads to certain purchases,” says Rudi, noting “It’s just as important to understand the approach of the people who make large purchases as it is to understand the behaviour of the people who don’t end up buying.”
All information is logged into a computer and aggregated to offer retailers statistical and contextual information they can use to develop their own marketing strategies, understand purchasing cycles, predict trends and manage inventory. “Computers are good at finding trends, at looking at things through ten dimensions, while it’s very difficult for people to relate to three dimensions at once,” says Rudi.
The analysis to date has produced some interesting results.
“The company was surprised at the impact that specifics like group size, gender and time spent in the store had on purchases and the differences between stores.”
The lowest conversion (sales per person) rate recorded was ten percent, the highest was 36 percent.
One store for example, showed that on average nine percent of single people who enter the store for one or two minutes will make a purchase. If they stay between two and four minutes then the chance of a sale goes up to 21 percent. Between four and eight minutes there’s 38 percent chance they will buy something. A group of two or three raises the conversion rate significantly but in a group of four or five the increase is minimal.
And then there’s the gender difference. The same store showed a conversion rate amongst single female shoppers of 22 percent, while the rate among single male shoppers was just ten percent. Surprisingly more males per group meant significantly more spending. However the more males in the group, the less time they spent inside the store, particularly as the group got larger.
Stores can use this sort of information to make decisions on matters such as how to cater for the bored male so he doesn’t make the group leave early. It could simply be a matter of bringing in a chair or a television or having an iPad app and some iPads.
Hunters and farmers
Tracking customers as they shop can help with designing store layout and improving product choices and placement while analysing sales techniques can identify best practice of sales staff.
“We note the salespeople who are good ‘hunters’ that is they are able to persuade customers to make a purchase and the ‘farmers’, staff who can convince people to upgrade or expand their purchases,” says Rudi.
This can assist with training sales teams to make better judgments on whom to approach, when and how often. Decisions can also be made on staff numbers at particular times and employee placement based on who works well together.
Data will also be compared to information gleaned from exit interviews and loyalty cards and scrutinised for repeated patterns.
“Since different stores and locations cater to different demographics of customer, you can use this type of data and analytics to understand how that manifests itself into shopping behaviour and get a better understanding of why some stores are more successful than others, which will help in choosing new store location for expansion,” notes Rudi.
Continuous observation will lead to an ongoing cycle of improvement, where retailers can analyse, influence, and then observe again.
Analysing behaviour not people
Similar tracking can be used to help other service organisations such as hotels, hospitals and airports.
“Wherever you have behaviour for people, you can analyse it. Now that we have the process in place, it’s easy for us to ramp up.”
And how will privacy groups react to this roll-out of technology? “While this might seem to be intruding on privacy, it’s not,” insists Rudi. “There are clear signs in stores that cameras are installed. Once the data is coded, then the videos are deleted and the data becomes anonymous. We don’t analyse specific people, we analyse patterns of behaviour.”
Nils Rudi is a Professor of Technology and Operations Management at INSEAD.