Retail analytics: customer tracking

While online shops generate almost too much data, which they can analyze to learn a great deal about their customers, high-street retailers are in a much more difficult position. But by applying various technical approaches, physical retailers can also analyze customer behaviour in their stores.

Retail analytics may sound very modern, but at its core, it’s something that retailers (at least larger ones) have been doing regularly. It includes measuring the number of customers, gathering information on their satisfaction and observing them.

The traditional approach, which managed without involving any technology at all, was based on using staff or external personnel to count and survey customers. But this is time-consuming and therefore creates significant costs. In the case of surveys, much depends on the interviewer. And it takes time for reports to be generated from the results.

In this respect, modern technology can help by providing results considerably faster. It also provides the option to link the key figures obtained with other data, such as weather, hyper-local information on visitors to the city center, branch revenue and so on. But before analysis can be performed, the measurements need to be taken.

How data can be gathered at the POS

It makes sense to use technology that already exists, For example, most high-street retailers have probably already installed cameras on their premises. Thanks to AI and advances in image recognition, software elements from surveillance suppliers can provide very good results on customer numbers, where customers walk and their dwell time. But monitoring every inch of larger stores requires a corresponding investment. There are also data protection issues to contend with when detecting and visually tracking people for analysis purposes.

As most customers already have a smartphone in their pockets, many vendors of analytics solutions use technology that identifies customers using their mobile devices. This often involves combining various approaches to capture as many customers as possible. WiFi tracking is combined with beacons, which use Bluetooth technology. A corresponding number of beacons (small radio modules) or WiFi sensors then need to be installed to monitor the POS.

For technical reasons, however, the results might not be clear-cut. The higher the number of Bluetooth receivers (i.e. customers) in the same room, the less precise the analysis. This applies to both time-based and spatial logging. And of course, any customers entering the shop who have switched off WiFI and Bluetooth, or who do not have smartphones, cannot be captured.

A completely different approach uses ultrasound. Implementing this technology is likely to be cheaper than installing beacons, due to the lower number of transmitters required. It involves transmitting signals inaudible to humans that are received by customers’ smartphones. For this to work, customers need to be persuaded to install the retailer’s app, which is used to perform the analysis. This also applies to analysis using a special frequency range emitted by the lighting.

Beacons, ultrasound or light can also be combined with an app for the purpose of navigation within the shop. The technology a retailer chooses to use depends on the retailer’s willingness to invest, structural possibilities and the analysis functions in the provider’s software, among other factors. There is no right or wrong approach.

The more detailed the information, the more complex the technology

The technologies outlined above are capable of providing some key parameters for managing a store, such as the number of customers, where they walk and dwell time. These parameters can be easily linked to other data, such as from the checkout system.

If you want to gain deeper insights, however, you will need additional specialized technology that is tailored to your particular industry. Let’s use fashion retailing as an example. The fitting rooms are central to whether customers decide to make a purchase. The following may therefore appeal to fashion retailers:

• Fitting room conversion rate: which products are particularly frequently tried on and then purchased? If the figure is low, the retailer may wish to reconsider how they position their products. They could also place product notices directly by the fitting rooms or have their staff refer customers to specific items.
• Cross-selling rate: in fashion retail, the combination of several items is a key figure. If someone buys a shirt, they will likely want a tie to go with it. And how about matching a pair of trousers with a jacket?

Obviously, these numbers have to be determined differently. “Smart” fitting rooms are the answer. These use RFID labels, which are attached to the clothing, to discreetly find out which items the customer tries on. This digital technology can also recommend other products on display screens. The Otto Group in Germany shows what is already possible with this technology in its Fashion Connect Store, which was recently won an award for being Store of the Year.

Retailers need to do the necessary groundwork

As we can see, there is a whole range of different approaches and procedures to find out more about customers. But the same maxim applies to high-street stores as does to online shops: before starting to analyze, retailers have to know what they want to find out. At the end of the day, the numbers can only answer the questions which have been asked.