All retail environments, both physical and online, are concerned with understanding what customers do inside their stores. For bricks-and-mortar stores the questions may include: at what time do customers come in? Do they rapidly dart in and out seeking specific items? (E.g. milk or bread.) Do they linger, exploring different sections? (E.g. shoes, lingerie, cosmetics and the latest jeans.) Do they notice and stand in front of new displays? Do they pay attention to endcaps? Do they appear to scan various sections in search of something? What percentage of them go through the checkout (a conversion)? Or simply exit the store without buying anything? Are visitors’ paths through the store similar? Are “shallow” visits prevalent? And if so, why? Does it happen that potential customers walking in the store take a peek at the length of the checkout lines and leave because they are too long? What times during the day/week does this happen? And by the way, are these new or repeat customers? How often do they return? At what times? Do they come on Saturdays? Or right after work throughout the week? It’s a long list.
Data analysis enables physical store managers to
- segment their customers according to buying behavior, such as time of day or items basket
- learn where customers spend their time: they enter the store, visit specific departments, given aisles seeking product categories, then looking at specific items before purchasing (or not)
- understand what sections of the store are less visited, to inform layout changes
- learn the effectiveness of their cash/wrap section at busy times by measuring queue lengths
- assess the success of in-store promotions and marketing efforts
- measure the impact on sales of long lines at the registers during the busiest times of day.
Brands also benefit greatly from this data. Every brand tries to stand out. The best placement is often off-the-shelf, on a dedicated display or an endcap where the brand and its products are showcased. How many customers actually seek it? How many stop by? How many others looking for other products in the same category pay attention to it? Is there correlation between the number of interested visitors and increased sales? Most definitely a brand will avoid being displayed in a section with few visitors, or with plenty of visitors at the wrong time of day (e.g. dinner foodstuff when visitors come in for breakfast items). Clearly, analytics based on measuring visitors behavior can be very relevant for product placement, promotions and marketing campaigns, most of which often command large budgets.
How is data collected?
Achieving the greatest insights requires following the average customers at deeper levels of granularity, from the store, down to section, aisle, category and specific products within it. The traditional way to do this relies on trained observers (people) deployed through the store, discreetly observing customers’ behavior, recording their observations, often aided by video cameras.
Enter location technology-based analytics, enabled by the proliferation of smart phones: we can track people’s whereabouts by determining the location of their mobile devices over time.
Collecting device location data for retail analytics depends on several key factors:
- the ability to track visitors unobtrusively, without requiring people to connect to any system or activate mobile apps on their devices
- the ability to track all device brands
- obtaining location accuracy with granularity down to the aisle level, covering multiple floors and buildings, if necessary
Why location accuracy matters
Detecting and triangulating a device’s position by sensing the signals emitted by its Wi-Fi radio is a very practical approach, given the widespread availability and use of Wi-Fi services inside buildings. Also, many people never turn off their phones’ Wi-Fi radio. A location accuracy of about 2 or 3 meters is possible, enabling aisle-level granularity.
The question is often asked as to whether it is possible to use a store’s existing Wi-Fi infrastructure to collect the data. Clearly, it is possible to collect some data, but unfortunately, location accuracy will drop down to 10-12 meters of worse, making aisle-level granularity impossible. The reason why is that existing Wi-Fi systems are deployed for connectivity, not for location. Access points located in the center of rooms or sparsely distributed through a space make it nearly impossible to triangulate devices’ location with acceptable accuracy. Collecting section-level visitor counts is possible, but aisle-level location is not.
Collecting the right data is only the beginning. It must be aggregated, processed and presented to deliver the desired insights. Navizon customers provide several great examples.
They deliver real-time in-store shopper behavior analytics, enabling stores to get to know their customers by providing store and brand managers with actionable insights. The systems being deployed at some of the largest retailers in North America, South America, Europe and Asia leverage Wi-Fi technology to deliver deeper analytics to retailers, helping them improve their store execution performance and merchandizing effectiveness.
They provide real-time, in-store shopper behavior analytics with aisle-level accuracy and seamless POS integration. This lets retailers follow their customers through their in-store experience: where they went inside the store, how much time they spent in different areas of the store, observe what captured their interest and ultimately what they puchased. This type of analytics, which has been available for online shopping, is now being brought to the bricks-and-mortar world powered by Wi-Fi technology.
Behind the scenes
Navizon’s hardware-based products enable automatically compiling the raw data to deliver all these insights. Navizon I.T.S. (Indoor Triangulation System) detects the Wi-Fi signals emitted by devices like smart phones and tablets, determining their location as they move about, or their proximity to a desired spot, or just counting their numbers.
Privacy is always a concern. However, note that it is impossible to identify who the people are or obtain their phone numbers. No personally identifiable information is ever collected. Only device’s unique identifiers, their MAC addresses, which can be easily hashed or scrambled for greater privacy protection.
To collect location data, sensors are placed throughout a store (e.g. in every corner), to detect mobile phones whose Wi-Fi radios are active. The system estimates in real time the sequence of locations as mobile devices move through the site. Average location accuracy of about 2 or 3 meters is sufficient to know whether the device is on a specific aisle, standing in front of an endcap or queuing at the checkout line. A new era of retail analytics is now possible.