How is AI and IoT Changing the Retail Industry? (Part 2)

How is AI and IoT Changing the Retail Industry? (Part 2)

We previously wrote about how AI and IoT are transforming retail analytics. We discussed how new analytics options are being made available to physical stores. These give them the ability to be more like e-commerce stores, with the capacity to A/B test and provide real-time recommendations in physical space. In this blog we will discuss even more ways that these retail IoT, AI and Cognitive Services are changing the physical retail space. These include new ways of managing queues, social distancing and on-shelf stock as well as expanded options for determining price.

How is Retail Iot Changing Queue Management?

Queues are still a tedious reality of many public and private service providers. Queue management technology is an unglamorous but important field, especially as population growth increases demand on services. Queue management can be explicitly tied to efficient daily operations, since deciding how many tellers/checkouts to have open at any given time directly impacts KPIs and revenue. 

Research has shown that, on average, people are only willing to wait slightly more than two minutes in a queue that is not making progress. Ticketing systems have been implementing RFID technology and providing real-time updates to customers via their smartphones. Situations in which ticketing systems are not suitable, however, still have the potential to evolve through the implementation of retail IoT sensors and machine vision.

By being able to automatically track queues at the level of the individual using retail machine vision a dedicated system is able to see when a customer enters the queue and how long they have been waiting. It can detect if the wait time is approaching a previously calculated queue abandonment threshold and automatically send a notification to a manager. At this point a new checkout may be opened or other measures may be taken to prevent the customer attrition which queue drop-off represents.

a grocery store set up with reail ai and retail iot sensors

IoT Retail Applications for Enforcing Social Distancing

Following all of the government-enforced restrictions on gatherings and distancing during the time of COVID-19, physical retailers have been some of the most affected. In many cases, even as the restrictions are eased, distancing restrictions and occupancy restrictions add a layer of complexity to managing the reopening of stores. IoT systems that keep track of store occupancy and crowd movement for analytics reasons can be easily retooled in order to assist with COVID-19 related customer restrictions.

Real-time alerts can be sent to managers or floor staff if store occupancy increases above a designated level. Heat mapping tools can also be used to analyse which areas of the store are causing bottlenecks and clusters, leading to violations of distancing guidelines. This is especially useful for large retail store spaces in which it is more difficult for staff to keep track of the entirety of the space at all times.


How is Retail AI Changing Shelf Availability Management?

Stock replenishment in a retail space is typically not particularly responsive. Once the inventory management system indicates that the stock of a particular product is low, it is placed on order and the displays are replenished in waves as these orders come in. 

For a retailer, having the product be, not only in inventory, but also on the shelf at the time when it is needed is vital for retaining customer loyalty. A little over 30% of customers, not finding a product on the shelf, will opt to look for it at a different store, costing retailers almost 5% of their total annual sales. Even those that don’t will spend over 20% of their entire in-store time resolving these out of stock issues. 

The use of retail RFID technology to track stock replenishment is an established and widely used technology. These sensors would typically track pallets as they are delivered into the stock room and as these pallets are then taken out onto the shop floor. Since this data is already been gathered, the next step is to feed it into machine learning models. This opens up new ways to predict and anticipate demand, and be able to pre-empt shortages and bottlenecks.

RFID tracking methods, however, are not feasible to use for individual stock items on the shelves. This is because it may not be feasible to ensure that every product is properly tagged or report accurately on products that are unpacked into the wrong section. High fidelity retail IoT visual sensors combined with AI Machine Vision can solve this problem by being able to identify not only how much stock remains on a shelf, but whether or not it is of the appropriate type even if it is not optimally arranged. These kinds of sensor combined with real-time alerts open up the possibility for effective priority based-stock replenishment. 

a top down view of an iot and ai augmented retail space

How is Retail AI Changing Pricing Strategies?

One of the main levers that is available to sellers operating both within e-commerce and traditional retail is pricing. Understanding how price changes effect buying patterns is one of the most fundamental aspects of retail strategy. Although development in this field have been very much trailblazed by the big online retailers, there is no reason why physical retailers can’t implement the same machine learning based analytics to guide their pricing strategies. 

Determining an optimal price is often not a straightforward process and there is no single best formula to use. The effect of concurrent promotions, sales cannibalisation, cross-selling, seasonal and geographical variability and many more play into the decision making process. There are too many factors for a human to take into account and pricing is frequently performed based on instinct and an overemphasis on less relevant indicators such as current stock volume. 

Given the large number of factors that have to be considered, pricing is an area which benefits a lot from the application of machine learning models. Something which is potentially more of a challenge for physical retailers than their e-commerce counterparts is ensuring that all the relevant data points are consolidated and available in the appropriate format. Once this is done, the application of machine learning to pricing has show to be able to achieve a 15%-25% boost in item sales.

If you’d like to know more about how these new technologies are transforming traditional storefront-based retailers, watch the free webinar by BizData which covers a range of real-world examples.