Customer Behavior in Retail

Customer Behavior in Retail

In an industry known for its intense competition. Personalization of a shopping experience can drive a 40% larger basket (transaction size) according to the Boston Consulting Group. But, how can retailers better understand their customers, particularly in-store, in order to personalize the experience?  

From the arrival of the first online retailers, and the eCommerce channel many have established, the retailer has maintained visibility of how customers behave on their site. Behaviors like how they get there, how they navigate the site, what they look at and for how long, what they put in their basket or save for later, what they buy or where they abandon a cart, and whether targeted promotions or suggestions influence each of those decision points all inform a retailer’s perspective.  

The same is not true of physical stores. The people counters at the entrance are notoriously inaccurate and the only other reliable point of contact is frequently at the Point of Sale (POS). The reality is: retailers only know roughly how many potential customers entered the store, how many transactions took place, and what was purchased. Very little other information is available. What path did a customer take? Did certain displays attract them? Where did they pause (or dwell)? Did they interact with a fixture, product or associate? Did they pick a product up, consider it, and put it back, or put it in their basket? Could an obstruction in an aisle, such as a cart, display, or associate stocking shelves impact sales of a particular product?  

These are all things we could see if we were to follow a customer around, but that would be kind of creepy. Historically, the best retailers could hope for a survey completion from their customers, either from a random intercept by an intern with a clipboard, or in the vain hope that someone completes an online survey using a link that the cashier circles on your mile long receipt. The typical take-rate for those surveys is around 1%, which is statistically insignificant. 

Today’s brick and mortar stores 

Retailers are trying to understand customer behavior, not only to deliver a personalized experience to grab additional consumer wallet share, but also to allow them to do more with less staff. Understanding that behavior would also help them optimize stock levels and product assortment based on historical and predicted demand, as they can with the online presence. While suffering an out of stock is a cardinal sin, holding too much stock represents an investment tied up in product that may end up needing to be discounted if it does not move fast enough.  

Retailers refer to “fast movers” and “slow movers” that represent the velocity of stock turnover. Fast movers sell out quickly and may need regular restocking. Slow movers need a lower stock level and if sales stagnate, may need to be promoted or discounted to move the stock.  

Does the retailer have hot spots in the store that suffer from congestion that may cause customers to avoid that area? Are there cold spots where the traffic is light and products in that area are slow movers?  

Do displays, such as endcaps and printed signage influence the flow of traffic in the store?  

If the retailer implements dynamic digital signage that responds to traffic flow, specific promotions, time of day or demographics, can they drive buying decisions in the store?  

In many cases Consumer Packaged Goods (CPG) manufacturers pay for their merchandising location in the aisle, think of Coke and Pepsi in the beverage aisle. The position is not accidental, they pay for that spot. To justify any costs to the CPGs, the retailer must be able to show information on brand or product impressions.  

What’s the solution? 

By leveraging Smart Cameras with advanced AI models, also referred to as Computer Vision (CV), a retailer can “see” all the behaviors that eCommerce takes for granted. They can obtain accurate counts of people entering the store, with the potential to exclude staff, delivery drivers, and others who are not potential customers from the count. In addition, the AI models can provide demographic information on the customers that may shape their in-store experience, e.g. They may determine that females between 35-45 shop more on weeknights between 7-9pm, hence is there an appropriate action they can take to personalize the experience for them?  

Similarly, the retailer can determine where customers dwell to look at a product or display and whether they interact or engage with the display. Do they pick up an item, consider it and return it to the shelf? Or do they go on to put it in their basket? If they take a long time to evaluate the item, can the retailer incentivize them to purchase the product through a promotion sent to nearby digital signage, the consumer’s mobile phone, or an Electronic Shelf Label (ESL)?  

Of course, we have the problems of lines or queues, whether at a service counter, the checkout, or increasingly, for curbside pickup or at a drive-thru. Excessive wait time can cause consumers to abandon their transaction, referred to as balk. This results in direct loss of revenue for a retailer and in fact may add cost to restock or through wastage for perishable items. Using the VIA approach (Visibility, Insights, and Action), the retailer can detect, or better yet, predict, line growth. They can understand whether the condition is transient, will persist or worsen. The retailer can then take action to resolve the situation before it becomes an issue. As an example, at a Food Service drive-through, the line is growing, the AI knows that at six-cars deep the average wait time at this time of day will be ten minutes. At that point, cars will start to pull out of line (or balk), so the system starts to promote low and no-prep time items on the menu board to reduce ticket-time and hence shorten the line.  

“My retailer says they already have cameras.” This may be true, but most cameras in retail are for loss prevention (LP) only and are only capable of recording video. Some video analytics companies will tap into those feeds and perform analysis using an on-premise server or in the cloud, however LP cameras generally are not positioned to capture what we need for behavior. Plus, few retailers want additional servers in the closet or traffic on their network. The Meraki MV smart cameras allow for edge processing of AI models reducing the network traffic required and reducing the need for larger on-premise servers. They also allow for sophisticated LP models that detect loss, rather than just recording it. 

Leveraging Meraki MV cameras 

By leveraging Meraki MV cameras and partner AI models to better understand consumer behavior in the physical stores, retailers can:  

  • Understand how customers engage with the store, where they go, where they dwell and interact.  
  • Personalize shopping experiences and deliver those experiences to the customer’s device or via digital media in-store, which drives incremental revenue and loyalty.  
  • Optimize the product assortment and stock levels to reduce investment in stock-on-hand and reduce the risk of out-of-stock.  
  • Generate incremental revenue through paid merchandizing.  
  • Manage consumer wait times to minimize balk and capture revenue that may otherwise be lost. 

Cisco not only understands this new world of retail, our solutions make it possible.

Share: