Digital signage is continuing its slow, steady growth across all sectors around the world. Accompanying that growth is the increasing challenge of delivering relevant messaging for the unique viewers watching the screen.
Our web browsers and smartphones do an excellent job of keeping track of what we do. One primary reason they track us is so that marketing messages and advertisements are more relevant. We’re more likely to pay attention to the ad, and possibly purchase the product or service, if it’s targeted directly to us. With online advertising, the ads are specific to what we’ve searched for, our online activities and the places we’ve visited in real life. Depending on your smartphone’s privacy settings, social-media, shopping and gaming apps are quietly gathering details of your life so as better to tailor messaging, thereby delivering a better return for the advertisers who support these “free” services. The technology includes immediate, cloud-based artificial intelligence (AI), and it has gotten so good that we wonder if our devices are secretly spying on us or listening in without our consent.
Digital signs don’t have a personal profile of each viewer, and, in light of recent privacy laws, such as the European Union’s General Data Protection Regulation (GDPR), they probably never will. Without that same level of individual tracking, digital signage is limited to personalizing messages for viewers using demographic targeting. That’s done using video analytics, a combination of cameras and on-premise software to determine the age, gender and emotions of the person viewing the screen. With those basic facts, the digital sign can update messages immediately and deliver a more relevant advertisement for the unique individual looking at the screen.
An example that’s easy to understand is a big-box clothing store with digital signage at the entrance. Each sign is running a loop of predetermined ads based on aggregate shopping data, regional demographics and seasonal shopping patterns. Because most transactions are card-based, the retailer has some facts about its shoppers, and it uses educated guesses to deliver ads that are relevant to the largest segment of those shoppers.
Adding video analytics and programmatic content, the screen can change based on the viewer who’s there at that moment, showing clothes for younger women when it detects a 20-something female and showing items for mature men when it detects a 50-something male. Video analytics also includes real-time store metrics like people counting and live demographics to help fine-tune store advertising. This fine-tuning increases the relevancy of the screen content, potentially translating to improved sales.
How It Works
In modern retail, a combination of hardware and software is used to target digital-signage messaging and deliver the best return at the cash register. Video analytics and Wi-Fi tracking are combined with point-of-purchase (PoP) data, business intelligence and machine-learning software to develop a same-visit profile for shoppers.
The primary technology being deployed centers on facial-detection algorithms, which can find multiple faces in images, and AI to determine the demographics of each individual.
Video analytics uses facial detection only—not facial-recognition technology that’s intended to find and identify a specific individual. For example, a casino might use facial recognition to spot a VIP or a card counter, allowing guest services to lead one to a high-rollers table and the other to the nearest exit.
Facial detection is anonymous, and it’s used for demographics and advertisement targeting only. Facial recognition uses a database of known faces, and it might be cloud-based; conversely, facial detection typically runs on-premise, it doesn’t use a database and it doesn’t upload any image data to the cloud. Once a face is detected and analyzed, all that’s retained is aggregate demographic data—not images or video of the specific person. Recent privacy laws prevent the storage of the individual’s photo, and, typically, data is only used in real time for the same visit to the store. There’s no attempt to recognize the person’s face and compare it to social-media photos or connect to his or her online profile, for example.
Often, Wi-Fi tracking is combined with video analytics to build a profile of the shopper’s visit. As someone walks into a retail store, his or her mobile device’s Wi-Fi is trying to connect, but the store doesn’t know exactly who he or she is. That person’s smartphone is randomizing information to help obscure his or her identity. Even without a unique signature of who that person is, though, Wi-Fi tracking can determine that individual’s path through a store, the aisles visited and the length of the shopping trip. Because there’s no way to know exactly who the person is, the information is only used for check-out-signage messaging and business intelligence.
Wi-Fi tracking combined with video analytics builds an anonymous same-visit profile of activities to help target messaging more precisely for future customers. When purchase data is tied in, the retailer has better business intelligence to build the best possible experience for shoppers.
Modern retailers are using this data so as better to stock their shelves and plan their displays/store layout. Analyzing traffic patterns by time of day, seasons and other factors can help find hot and cold zones within a store. Retailers are learning about high dwell areas, where shoppers tend to linger. They’re also using this data to group together products in ways of which they’d never previously thought. If many customers buy product A and then walk halfway across the store to buy product B, then those products can be grouped together on temporary displays to improve sell-through.
Video analytics and Wi-Fi tracking also help with employee resource planning. In stores that have these technologies, employees don’t have to wait for lines to form at the register. They receive audio cues over in-store speakers or radio buzzes to alert them that large groups of shoppers are about to finish shopping and check out, even before the customers have arrived at the register!
Video analytics also means more cameras in store, which can be integrated into loss-prevention and security systems so as better to protect profits and prices.
Using these technologies, retail stores can compete with online outlets by delivering a better overall customer experience. Online is convenient, but in-store technologies are helping reduce common complaints about brick and mortar. In general, customers like to interact before purchasing, but they also want a “frictionless” overall experience. In other words, they want both discovery and convenience.
Video analytics isn’t just for shopping. It’s used in banking, hospitality, healthcare and event spaces. Knowing your viewer means delivering relevant messaging anywhere—and not just to get someone to buy something, either. In healthcare, waiting-room signage that can determine who’s in the room and watching the sign can display age- and gender-specific educational messages.\ Video analytics can improve relevance in any situation in which a wide variety of visitors is present and there’s unique content to share, depending on the exact viewer in front of the screen.
Although facial detection is not as individually targeted as facial recognition is, there are still some legal concerns regarding privacy. The EU’s GDPR regulations took effect in summer 2018. The California Consumer Privacy Act (CCPA) regulations, which are similar, became ef fective in Januar y. These laws mostly cover online privacy and the ownership and usage of personal data, but they also af fect video analytics and surveillance systems. For camera systems, one main requirement is simply to inform visitors that cameras are in use and protect the video data from hacking and breaches. Major video-analytics-technology providers approach these current regulations by not storing image data and through on-premise data processing, as a result of which these privacy laws don’t apply in the first place.
As clients begin to request these technologies, it’s critical to have expert resources on hand to help start the process. The top digital-signage content-management systems support video-analytics software, whether it’s directly built into the software or whether it’s through partnerships. It’s important to request case studies and review real-world examples of these systems at work. Larger national retailers are developing their own systems, but medium-sized ones rely on integrators to assist them in putting together the right combination of technologies to deliver on their goals. Large industry trade shows have sessions that relate to video analytics, and the Digital Signage Federation (DSF) has an online certification course specifically for these technologies.
There’s so much more to these technologies—in particular, the unique ways to create digital-signage content for video-analytics systems. Next month, we will explore advanced methods of creating content for these systems.
Thoughts From A Category Expert
“It is important to understand the differences between facial recognition and detection. Facial recognition actually captures an image of a person’s face, whereas facial detection is the result of a calculation applied to the geometry of a person’s face. The aggregation of that calculation assigns an age and gender. No image is captured or saved. Facial detection is anonymous, whereas facial recognition captures personally identifiable information.”
“Wi-Fi sensors are used for people counting so businesses can understand how many unique visitors have visited different areas of their business, how long they are spending in different areas and, in some cases, where they are traveling in the environment. This is used for analytics, for store planning and to live trigger digital content.”
—Kelly Harlin, Analytics Platform Strategist, NEC Display Solutions of America, Inc.