How to get into the head of a visitor, or about Wi-Fi analytics

    Imagine, we can analyze the way customers with grocery carts: how and where they moved. Based on this, you can improve merchandising, explore which products from which groups are buying together, and much more. But the most interesting thing begins when a person already comes to the cashier: the time spent in the store, the route and its final purchase can be synchronized, and the customer organization can also combine this information with its loyalty program.

    One of the largest developers of Wi-Fi analytics system in Russia, used in retail, shopping centers and HoReCa facilities, shares its secrets.

    Offline metrics

    Any modern phone has a Wi-Fi module, and if it is turned on (and usually it is turned on), then without the owner’s knowledge it starts sending multiple signals. These guys “catch” signals using TP-Link equipment and their firmware. The signal contains the MAC address of the phone, the power with which the signal was sent, and the time it was sent. There are over 15 different offline metrics that can be calculated based on this data. Here are some of them:

    • Entry rate - a metric showing how many people passed the location and how many of them entered it. Entry rate allows you to evaluate the location potential, the volume of the audience with which the business works, as well as analyze the effectiveness of a particular marketing activity. Due to the fact that the MAC address is an individual identifier of the phone and, as a rule, it is not repeated, it is used to evaluate such parameters as “new visitor”, “repeat visitor”, etc.
    • Frequency is an important parameter especially for retail. This metric shows the frequency with which new or repeat visitors come to a particular object, you can see the cycle of repeated visits.
    • Routes . If there are several data collection points, you can analyze the routes of visitors, the proportion of people entering certain rooms (shops), etc. You can unite people in groups and analyze how they, having visited tenant X, went to tenant Y, went to the cinema and so on - and based on this information develop a loyalty program.

    Now many are trying to work with a loyal mass of audience, because the purchasing power of the population is declining. In shopping centers, for example, 80-85% loyal audience, because visitors are mainly residents of nearby houses or office workers. Therefore, shopping centers try to attract them, reward for visits and stimulate purchases.

    • Average time . This is what affects our purchases, because the more time we spend in the store, the higher the likelihood that we will buy something. It is generally difficult to leave a shopping center without a purchase: there is more opportunity to spend a huge amount of time and, accordingly, to acquire something, go to the movies, etc.
    • Intersections are another very interesting metric for business. This metric shows where visitors of one location still go inside the eco-system of Wi-Fi analytics, including amusement parks, sports facilities. These data, for example, allow a company to better understand the behavioral portrait of its consumer.

    Operating modes

    TP-Link equipment can operate in various modes.

    • Providing Internet access for staff and visitors

    We promote Wi-Fi not only as a service for visitors, but also as a profitable resource for business and the customer. Now Internet access for visitors is a loss-making article for many companies. We see the opportunity to make Wi-Fi for visitors a profitable article.

    • Loyalty Program Recruitment Tool

    Wi-Fi is a way to communicate with a client. Our project partner Shopster comments: “There are many ways to communicate with a client. For example, a user who connected to guest Wi-Fi saw a banner / video, etc. - this is already communication. But we go even further, create complex integration systems with the customer. Wi-Fi can be a recruiting tool for a loyalty program. ”

    For example, you can make a connection between the Wi-Fi analytics system and the customer’s CRM system and, when the visitor connects to the network, run several scripts.  

    One such scenario is when a user reconnects to a customer’s network — for example, in another coffee shop or restaurant on the same network. According to the law on the provision of public Wi-Fi networks, when connecting, the user must leave contact information and confirm a phone number. Since the system is unified, the first time you connect, the user's MAC address and phone number are stored in a common database. And the next time the same user uses Wi-Fi in another coffee shop on the network, he will not have to re-login.

    Another scenario - at the moment when the user connects to the customer’s network, where Wi-Fi is used as a recruiting tool in the loyalty program, the CRM system has already been asked whether this user is a member of your loyalty program. If not, then the user is shown a questionnaire or any other form where you need to fill out only partial information (for example, name, gender, age), and the phone number is already there. He can only confirm - and the profile immediately flies to the CRM-system, and on your smartphone or tablet the user’s account immediately opens. Such a scenario facilitates the interaction and increases the likelihood that the user becomes a member of the customer loyalty program.

    One big customer of Shopster analytics Wi-Fi system is actively using this tool, but in this case, it’s not the recruitment into the loyalty program, but the following.

    Trigger platform

    The synergy between the analytics product and the hotspot product is important. When there is a bunch of MAC address + phone number, then the system can work as a trigger platform . For example, take a jewelry store. We see that the user passes by a store with jewelry. We transfer this information to the customer’s CRM system in real time, and she sees that such a number is in the database and he often buys jewelry. The customer understands that he can communicate with this user: send SMS, for example, about available bonuses or a notification about a closed sale. As a result, a potential buyer receives an interesting offer exactly when he passes by the customer's window. It turns out such a bunch of loyalty programs, analytics and hotspot.

    In a typical scenario, SMS messages with stocks almost always arrive at the wrong time. For example, you are at home, and some well-known sports brand sends an SMS that 1000 bonuses have accumulated in your account and you need to spend them until the end of the weekend. It is unlikely that it will encourage you to pack up, go to the store and buy something.

    Shopster's idea is to do what is now called Super Geo Communication in the world of internet marketing. That is, the Wi-Fi analytics system works as a trigger - it “catches” the user's MAC address and phone number and transfers it to the customer’s system. The latter (if he has permission) builds communication with the user: sends SMS at the right time.

    Audience extension tool

    More interesting. The company Maxima Telecom, which provides Wi-Fi in the metro, has a database of over 19 million bundles of MAC addresses with phone numbers. Using this database, you can expand the scope and work as a trigger.

    How it works?

    TP-Link equipment with Shopster firmware installed in various stores and shopping centers catches the MAC addresses, which are then transmitted to the Maxim Telecom system, where it recognizes by MAC address which phone number it belongs to. The text of the SMS message pre-generated by the customer on behalf of is sent to the person who at some point passes by the store. The likelihood of “hooking” a person passing by increases significantly.

    This process is called audience expansion, and it is a new tool that Maxima Telecom is launching now with Shopster.

    Offline combination tool

    Another new product that Shopster is launching today with TP-Link and large customers is the online-offline combining tool. TP-Link equipment along with Shopster firmware has become a bridge connecting the physical world and the Internet world. It's no secret that on the Internet all sites monitor users, they put cookies on them, and Shopster catches their MAC addresses. So, Shopster learned to combine Cookies with MAC addresses: Shopster transfers them to large Internet sites, and recognition takes place on their side. Here is what it gives:

    • You can create a clearer picture of the interests and not only of your target audience.
    • You can evaluate the effectiveness of investment in advertising (contextual advertising, banners, etc.) - for example, conversion from advertising on the Internet. Previously, classic offline business was quite difficult to do. Now Shopster’s Wi-Fi analytics can help - by collecting MAC addresses offline, for example, you can identify how many people who viewed ads online, visited an offline location, and much more.
    • Also, this data can be used for retargeting - you can collect the audience that came to a particular store in one segment and provide it to the customer. After that, the customer can very targeted or “Super Geo” send online communication to these users, lure them, stimulate their return and thus develop loyalty. This is a new product that many customers are starting to use.

    Another product at the junction of network equipment, software and Bluetooth technology: a Bluetooth adapter with modified firmware is inserted into the device and it becomes possible to work with iBeacon or Eddystone, which provides fairly accurate indoor navigation (Indoor navigation).

    Indoor navigation can exist as an independent product for visitors (such navigation already works in one of the central shopping centers in Moscow (Shopster client)) or as an auxiliary solution for business (for example, staff tracking).

    Personnel tracking is a narrow retail need. In retail, the consulting model is in demand, that is, in the store, the consultant must necessarily communicate with the visitor to increase the likelihood of a purchase. Therefore, it is very important that the consultants are in the sales area during work, and not, say, in the utility rooms or in the warehouse. When implementing a tracking system, you will see where employees go, in which zones they are located. From this it is already possible to build statistics. And you can memorize the routes of visitors through grocery carts. At the checkout, the cart (and with it all the information collected) can be unambiguously “tied” to the customer number in the loyalty program, if he participates in it. This gives the store an even greater chance to please its regular customer and earn more.
    In this case, more serious mathematics is involved: Bayesian filters, a lot of linear algebra and machine learning.


    The Shopster analytics Wi-Fi system architecture consists of three layers:

    1. Infrastructure It collects and sends data. At this level, it is important to ensure equipment control, sufficient fault tolerance and monitoring.
    2. Data storage. Since we are talking about working with a large amount of information, it needs to be properly prepared and correctly stored - in order to quickly process it in the future.
    3. Business logic. Since the system has high natural complexity, it is important to be able to correctly calculate complex analytical metrics based on stored data.

    Now is the time to go through the main components of the system.

    Server part of the solution

    The server part is a collection of data storage clusters, a server with preprocessing services, as well as individual tools that provide targeted functionality for each part of the solution. In this case, both virtual and physical servers are used in several independent data centers (chipcore (selectel), firstDEDIC, firstVDS).

    This distribution of capacities is important because a huge amount of data needs to be processed.


    All solutions can be implemented on the basis of even one TP-Link device.

    In 95% of Shopster projects in retail, shopping centers and HoReCa facilities (which are about 2500 devices in Moscow, St. Petersburg, million-plus cities and other small cities) TP-Link equipment is used.

    “The TP-Link equipment that we use has features that allow it to work in all modes without failures and stably. Now we are working on the TP-Link EAP115 , EAP110 models of the Auranet line. Outdoor solutions CPE210 v.1 and EAP110-Outdoor are tested .

    TP-Link Ceiling Wi-Fi access point, Auranet series, EAP115

    TP-Link 9 dBi outdoor Wi-Fi access point, Pharos line, CPE210

    TP-Link equipment has a good service life - it works stably for a long time: the vast majority locations there has not been a single replacement of equipment since 2013, ”comments Ilya Spasenov, co-founder of Shopster.

    It is not that simple

    The system is based on a whole set of interesting algorithms, ranging from smart calibration of objects to data-mining algorithms. Even seemingly trivial tasks at the level of the formation and processing of technical logs in fact are not so simple. A lot of efforts of developers and hardware resources of devices are spent on the fight against noise and radio interference. About 60-70% of signals are garbage that will not participate in “useful” calculations.

    As described by Shopster: “Our work in Wi-Fi analytics sometimes resembles the work of a taxidermist (the person who creates stuffed animals), who was brought a rabbit carcass with a grenade. You are trying to make "beautiful" in some pieces.

    Instead of a conclusion

    Wi-Fi analytics and profit are synonyms. With the help of Wi-Fi analytics, you can get unique analytical data, unique sales tools. Moreover, it is inexpensive and applicable in different conditions.

    Also popular now: