AI of Things (VIII): socio-demographic segmentation and video analytics to improve shopping experience

Pablo Salinero    13 July, 2022
Photo: Blake Wisz / Unsplash

The growth of e-commerce, with the many advantages and conveniences it offers to customers, has meant that physical shops have seen their market share shrink significantly. More recently still, the forced closure of physical shops due to the COVID epidemic has forced traditional retailers to completely reinvent themselves in order to attract customers again.

This reinvention should be based on two main pillars:

  • maintain the differential offer of physical shops (personalised physical service, physical access to the product);
  • convert the shopping experience into something more similar to that of online shops (product recommendations, personalised campaigns, etc).

In other words, the physical shop must stop merely being a warehouse for products and become a space for services available to the customer. And, to implement this second axis, the digitalisation of the physical shop is essential.

What can physical shops learn from online commerce?

What lessons can a physical shop learn from the way online shops relate to their customers? In online shops, from the moment the customer registers, the seller already knows which customer is buying and can have all their socio-demographic and economic information.

In online commerce, the customer’s consumption and browsing habits are also known: which products they have visited, how much time they have spent analysing and studying each product, which campaigns or suggested products they have shown interest in and which they have ignored, which days of the week or times of the year they look for and buy certain products.

How should a physical shop evolve to be able to make use of similar information to improve the customer shopping experience?

First of all, it is necessary to sensor the shop, which will allow the collection of data to be able to analyse customer behaviour inside the shop. The main types of sensors that can be installed are:

  • Video cameras: with software installed that allows facial recognition. The aim is not so much to uniquely identify customers entering the shop, but to be able to count how many people enter and leave, their gender and age range. To ensure customer privacy, this information is not recorded.
  • Bluetooth sensors (beacons): devices that are placed at different points in the shop to locate customers via their smartphones and communicate with them to inform them of the details of a product, show them offers associated with a specific point in the shop or last minute offers, find out about their movements around the shop and the busiest points, etc.
  • RFID (Radio Frequency Identification): tags that are placed on products and that significantly improve the possibilities of barcodes.
  • Interactive screens: distributed throughout the shop, with a dual purpose. On the one hand, to show customers personalised messages and, on the other, to collect feedback on the user’s shopping experience.

Although the exploitation of the information provided by these IoT sensors alone, without combining it with any other type of information, brings benefits that are more store-oriented than customer-oriented, it makes it possible to determine the ‘hot’ areas of the shop, to optimise its internal design by redistributing shelves and products, and to improve the management of queues at the checkout.

However, the information obtained in this way has the limitation that it does not distinguish between customers and considers them all, broadly speaking, as belonging to the same group.

Digitalisation of shops to improve customer experience

How can the benefits of digitalisation be increased for the shop, but also to improve the customer experience? By cross-referencing information from sensors with socio-demographic information about customers: their age group, gender, household size, economic group, place of residence and place of work.

For legal reasons, the information used must be aggregated so that customers cannot be uniquely identified, but despite this limitation, the cross-referenced and enriched information allows segmentation according to socio-demographic profile.

Cross-referenced and enriched information allows segmentation according to the socio-demographic profile of buyers

When the customer enters the shop, the bluetooth sensors detect their mobile devices, so the customer is identified. However, as mentioned above, this is not the final objective, but rather to cross-reference it with socio-demographic information to determine the segment of the population to which they belong.

This information is at a theoretical level, because at a practical level the cameras detect how many people belong to the group that has just entered, their sex and their ages, bearing in mind, once again, that specific customers are not identified but their socio-demographic segments. With this information provided by the cameras, it is possible to know whether it is a single person, a couple, a family with small children or older ones, a group of adults or teenagers, etc.

Customer knowledge in the physical shop

The physical shop is now on a par with the online shop in terms of customer knowledge, since, having assigned the customer to their socio-demographic segment, it can use everything it knows about the behaviour of similar groups of customers to offer a more personalised digital service, such as presenting campaigns and offering more specific products.

Moreover, the behaviour of these new groups inside the shop (obtained from the tracking of their path provided by the cameras or their use of the interactive screens), allows you to refine this knowledge of the customer, making segmentations of favourite products according to the customer’s socio-demographic group or, conversely, socio-demographic segmentations for each product or campaign.

Photo: Ashim D Silva
Photo: Ashim D Silva

It is possible to extract even more information: the information provided by the purchase tickets allows us to obtain a measure of the relevance of the products according to the interest they arouse during the purchase process and the conversion rate of this interest into a final sale. And the interactive screens located at the exit of the shop allow to collect the feedback of the shopping experience and to know if the actions launched on the customer have helped the final sale or not.

Mobility information: where do customers come from (and where do they go)?

In addition to purely socio-demographic information, there is another type of information that can be useful when combined with the digitisation of physical shops: mobility information, which refers to customers’ travel and movement habits.

This information is extracted from the millions of events recorded daily on mobile networks, always anonymised, extrapolated to the total population and also aggregated into socio-demographic characteristics.

With this mobility information, it is possible to know where the customers who come to the shop come from, how often, what days of the week and whether they do so because they live, work or are sightseeing in the area. This information was already very useful even before the shop was opened, as it was used to decide on the ideal locations for the shops, depending on the socio-demographic profile of the customers to be attracted, looking for the areas where these customers move around the most.


The digitalisation of shops, from the double point of view of physical infrastructure and customer information, brings benefits for both the shop and the customers.

Shops can organise themselves more efficiently, placing products and displaying campaigns in a way that attracts more attention, and they can find out what kind of customers are the most frequent visitors, how they behave and what products they prefer.

Meanwhile, customers see their shopping experience improved as they receive a much more personalised attention tailored to their profile. All of this makes the customer happier, and a happy customer spends more, which in turn increases the shop’s profitability and profit.

If you want to know more applications of the fusion of the Internet of Things and Artificial Intelligence, or what we call AI of Things, you can read other articles of the series:

Leave a Reply

Your email address will not be published.