A RetailWire mPaper.  Underwritten by IBM.

AI’s New Starring Role in Customer Engagement

Why artificial intelligence is becoming instrumental for unifying in-store and online experiences

72-ai-with-customer-feature-900x900-r1

Introduction: AI’s growing influence in retail

Imagine you could read, remember and compare every news story and social media stream in your market every day and extract useful trends and action steps. How would your decision-making change?

While this might be a fantasy scenario for a human being, it’s entirely plausible for computers programmed with artificial intelligence (AI). After decades of intense development, AI is coming into its own in the retail industry. It’s a technology poised to exercise tremendous influence on the way businesses operate. It has inspired retail practitioners to reimagine what their businesses can accomplish. Retailers are increasingly using AI to help understand their shoppers better, to present personalized online merchandising and offers, and to tune up merchandising and supply chain processes — to name just a few applications.

AI has several attributes that make it well-suited for the everyday decision making that characterizes modern retailing across physical, digital and mobile touch points. Among them are perfect factual recall, rapid processing of large and diverse data sets, and time-saving automation for frequent, repetitive actions.

For retailers and their suppliers, AI is already making a meaningful impact in several crucial domains, including customer and market insights, shopper experience and engagement; and supply chain and operations.

Customer and market insights

Merchants apply AI technology to manage data used to understand consumers, the overall market and their competition at a much more granular level. Retailers can use AI to mine both structured and unstructured data in pursuit of actionable intelligence. In general, AI can deliver better management and handling of content, including user-generated content and images from social media.

Shopper experience and engagement

AI technology can enhance several facets of the shopper experience. It can facilitate better shopper experiences by sensing and analyzing shopper interactions across physical and digital touch points. In the digital realm, these insights may be applied to personalize the presentation of merchandise choices based on past behavior of the individual and shoppers with similar traits. Using insights derived from the interplay of shopper interactions, AI can then assist merchants in creating and automating highly relevant personalized marketing. In the physical store setting, AI can be applied to help localize assortments.

Supply chain and operations

AI technology can help keep better tabs on retail and consumer packaged goods (CPG) supply chains. Whether tracking ingredients upstream or monitoring goods transported by delivery truck fleets, AI promises much greater visibility of the flow of goods and transactions that underpin the industry. These advancements are especially apt in store operations. In inventory management, for example, this level of visibility can be a boon to managers pursuing high service levels while keeping inventory costs within competitive bounds.

Back to top

Where AI fits in the cognitive technology universe

Artificial intelligence, or the simulation of human intelligence processes, is one of several types of “cognitive technologies” that today are making a meaningful impact for modern retailers. Others include:

Robotics The conception, design, manufacture and operation of robots.

Machine learning Systems with the ability to learn and improve without explicit instructions.

Natural language processing The ability to understand human speech as it is spoken.

Deep learning Machine learning with artificial neural network algorithms.

Predictive analytics Predicting outcomes using statistical algorithms and machine learning.

Recommendation engines Systems that analyze data and make suggestions based on what’s known about each user’s interests.

It is not unusual for these cognitive technologies to be used in combination — for example, retail service robots that can respond to human speech.

Back to top

Why digital retail made AI imperative for traditional retailers

The rapid leap from point of sale (POS) and syndicated data to web server logs in the mid-1990s triggered explosive consequences that are still being assimilated today.

Web commerce pioneers introduced a new kind of customer interaction in which every moment could be tracked as data. Conventional retailers could see the writing on the wall. Many responded to the dot.com onslaught of the late 1990s by launching their own online ventures. It wasn’t long before the industry dialog began to shift from physical-versus-digital, to exploring how the two realms might coexist under a common brand.

By 1999, pioneering multichannel retailers like JCPenney and Eddie Bauer began presenting tantalizing facts about their shoppers to industry analysts: Multichannel customers, who interacted with their stores, their catalogs and their web sites, were becoming their most valuable customers, and by a considerable margin.¹ It made sense. More touchpoints meant more opportunities to sell, and shoppers already loyal to their brands were more likely to try to buy via the new digital storefronts.

As retailers gained more comfort with digital commerce, competitive impetus led them to seek the same level of visibility into their brick-and-mortar environments that was commonplace in online stores. Today, retail brands see that to mold successful omnichannel operations, they must manage their assortments, inventories and pricing holistically. The ability to have consistent item visibility and service levels across all channels of interaction is now a requirement.

Digital competition has taught the retail industry that it’s not enough to gather sales data and review performance reports after the fact. Things move too fast in the omnichannel era. Operators must routinely apply more types of insights to deliver more personalized and satisfying customer experiences at every touch point, every time. Retailers have also come to understand that these intricate and frequent decisions defy human capabilities.

¹ JCPenney.com and EddieBauer.com analyst presentations (1999)

SOURCE: WEBINAR: TIPS TO IMPROVE THE CUSTOMER EXPERIENCE WITH DIGITAL INTELLIGENCE April 2016.

Back to top

AI and in-store sensing – a new insights explosion

To match the standards of analytical depth and responsiveness set by digital retailers, store operators have pursued analogous capabilities within physical stores.

Retailers are presently deploying various forms of in-store sensing in efforts to gain greater insights into shopper behaviors and interactions within the selling environment. Among the approaches:

  • Video analytics and shopper tracking
  • Bluetooth beacons
  • Low-power (NFC) communications with mobile devices
  • Monitoring frequent shopper card activity and promotional response

In some store environments, we also see the rise of RFID for tracking items, as well as near-field communication (NFC)-enabled product labels that allow shoppers to interact using their smartphones.Many of these devices leverage the Internet of Things (IoT) to connect the data they capture to various cloud-based analytics engines. The volume of information these systems generate can be massive as multiple sensors in multiple stores capture diverse data streams.

Comparable to the way digital retailers monitor the paths of online shoppers, store operators are using video sensing devices to learn about shoppers’ movement patterns and ”dwell times” on the selling floor. A few employ machine-based facial-recognition to classify shoppers or link them with prior visits.

These data may be aligned with transaction data from the POS or online platforms to better understand the performance of marketing, merchandising, messaging, store operations and promotional implementation.

For an omnichannel or unified commerce retailer, collating and interpreting shopper information from all these new in-store data sources on a continuous basis is not a trivial matter. But when accomplished, the results can paint a rich picture that rivals the sophistication of interaction data available from the online environment.

In turn, formulating responsive actions, whether broad-based or personalized, also becomes a massive and highly-detailed undertaking. Human decision makers need automated assistance to keep the pace. This is where AI assumes an essential role for today’s retailers.

Back to top

AI’s direct application to retail

At their best, artificial intelligences are a reflection of what customers do and how they think.

Natural language processing (NLP): Voice assistants like Alexa, Siri and Hey Google are transitioning from novelty to necessity, enabled by natural language processing. These services wouldn’t exist without AI, and machine learning also plays an essential role by continually refining the data view.

Deriving insights: Retailers can employ AI analysis and machine learning to derive useable insights from unstructured data, such as video images. Current examples include extracting shopper behavioral data from in-store video, interpreting facial recognition input and identifying products from images.

Pattern recognition: AI can memorize shopper interactions, driven by relentless analysis. When combined with machine learning, this type of system gains the reasoning ability to detect unanticipated patterns, trends and insights.

Shopper profiles: AI solutions can enable merchants to build a model of each shopper based on a combination of disparate data sources. An AI-powered, shopper-facing engine can put different interaction priorities into play at each moment of truth, based not only on past behavior of the individual shopper, but also on the behaviors of other shoppers with similar profiles.

Predictive analytics: Artificial intelligence systems can, in real-time, provide accurate item availability and delivery time by leveraging an intelligent forecast built from the store and fulfillment centers up. It can empower system-wide transparency about items and their locations, continuously updated to support error-free orders and optimal service levels.

Back to top

AI Case Study #1: Re-envision e-fulfillment

When a severe ice storm put 25 percent of its e-fulfillment capacity out of commission for 11 days, a 1,000-store apparel retailer suffered a 12-day order backlog that caused it to miss cutoff dates for pre-Christmas delivery. The combined cost of expedited shipments and other customer recompense exceeded $20M.

This retailer responded with an ambitious program to turn every facility within their system (e-fulfillment centers, distribution centers and stores) into a fully capable fulfillment location. This investment in resilience created its own set of challenges. For instance, the cost to pick, pack and ship an order in a store is markedly different than doing the same in a highly automated e-fulfillment facility. The challenge was how to create a system in which everything was in balance – inventory, shipping cost, handling cost, fulfillment capacity?

This was a multi-objective optimization process (MOOP) of the first order. It required an AI-assisted process to model a very complex set of business objectives. The inventory planners, logistics managers and store managers worked together to simulate the best outcome for the enterprise as a whole.

The solution yielded a hard dollar savings of over $7M (approximately 5 – 7 percent savings) in parcel shipping costs during the 11 days from Thanksgiving to the end of Cyber Week. At the same time the retailer was able to reduce average order-to-ship time by 30 percent and total order-to-delivery time by 33 percent over the previous year.

The retailer is learning to use these predictions to help balance demand as well as supply and reduce both markdowns and stock outs. AI is completely changing the way management looks at locating merchandise for its $3 billion dollar digital business.

Back to top

AI outside the store – the mobile experience

If only keeping abreast of the information from touchpoints within the sphere of retailer control was the entire challenge. Data flowing from a variety of external channels — notably social media and smartphone apps — presents another torrent of highly perishable data. Arguably, its volume, velocity and variety may exceed that generated within the retail domain itself.

Mobile now represents two out of three digital media minutes, and mobile apps are approaching 60 percent of total digital time spent, according to ComScore’s State of The U.S. Mobile Market 2016. The same report found that retail apps capture about 4 percent of interaction time spent by consumers, compared with 20 percent spent on social platforms and 16 percent on music apps.

Shopper behaviors within the physical and virtual spaces controlled by retailers are increasingly influenced by social media discussions, product reviews, how-tos and bloggers. Word of mouth has always been influential, but today no retail entity can afford to ignore the influences of two billion Facebook users worldwide on brand reputation and social sentiment. For many retailers, Pinterest’s 70 million monthly users have become a crucial target.

Retailers now look to their mobile channels for the dissemination of marketing messaging as well as vast flows of behavior and sentiment information. It is no longer optional to link individuals’ behaviors in the digital world to their in-store shopping actions. It is well documented that shoppers arrive at “shopping spaces” already pre-influenced by digital interactions with the category, the product and their peers.

For higher-consideration purchases, more shoppers now interact while they are present within the physical store space. They use mobile devices to compare prices, obtain product specs or ingredients, check shopping lists and download promotional offers on the fly.

Against this backdrop, retailers are now challenged with accounting for these outside influences when adapting their merchandise assortments, prices and personalized offers. Continuously sifting through the flood of data to separate relevant kernels of truth from extraneous information is a classic “signal to noise” challenge well-suited for AI systems.

Back to top

AI Case Study #2: Personalize assortments

An apparel retailer adopted an AI solution to understand the customer mission and create personalized shopping experiences. The retailer collates vast amounts of sales, loyalty, credit card, social media and browsing data using AI to read the “digital exhaust” that reveals each customer’s shopping preferences.

To tap more deeply into shoppers’ expressed likes and dislikes, the company added gamification. Customers interact with a Tinder-like game app that trains the AI and adds intelligence about color, style and fiber preferences. When a customer browses a category like women’s dresses (~3,500 products in all), the AI presents the most likely choices first, based on both history and expressed preferences.

Back to top

AI in omnichannel – an operational necessity

In this data-empowered marketing environment, personalization and tailored messaging have become integral to consumer shopping experiences. But further, the data that supports personalization is also the data on which retailers build their supply chain operations. These domains are interlinked and mutually dependent.

A highly responsive and efficient supply chain is one of the key drivers of the unified commerce. The online inventory visibility established by a handful of e-commerce powerhouses, like Amazon and Walmart, have set expectations for shoppers — online and in-store.

Ralph Jacobsen, Senior Executive of Global
Consumer Goods & Retail Marketing, IBM

“Machine learning is generating better insights for demand planning. I am seeing forecasting curves better matching actual demand curves, like never before. This is huge for retail operations, and is getting our industry closer to other industries that have the inherent benefits of knowing their incoming demand by having their customers make reservations, like hospitality and airlines.”

Store-level perpetual inventory and system-wide supply chain visibility enable omnichannel businesses to better predict demand and maximize item availability while keeping inventory levels within bounds. This in turn makes today’s research-anywhere, shop-anywhere, transact-anywhere, obtain-anywhere, return-anywhere service standard possible.

Meanwhile, new pressures from in-store picking and shipping of online orders, as well as buy online/pickup in-store (BOPIS) programs, add to the challenge of keeping items reliably in stock. Both pickup and home delivery transactions generate a wealth of shopper insights, while providing a highly-responsive and tailored experience to high-value customers.

AI can help retailers manage current inventory information with precision while also propagating crucial information to decision points further back along the supply chain. For physical stores, real time inventory accuracy and error-free replenishment orders can reduce delivery frequency and reduce safety stocks held at distribution centers. Retailers are also actively incorporating third-party source data, such as weather, news events and social trends, to sharpen the accuracy of supply chain forecasting.

AI can also be used to solve some everyday operational problems. Within stores, some of the same IoT sensing devices that are used to collect and interpret shopper behavior can also be used to keep tabs on checkout queues and advise staff on the right number of registers to keep open. Artificial intelligence can also help store retailers incorporate data about trip duration and movement into a type of predictive analytics that can help managers maintain high service standards while optimizing labor costs.

By leveraging the Internet of Things, AI can improve operating efficiency and extract costs from many areas of the business, while enabling more accurate and timely decisions.

Back to top

Following-up

About IBM Watson

Watson makes sense of the breadth and diversity of the world’s structured and unstructured data across a variety of industries, including sports, medicine, travel, retail and many others. For more information about IBM’s work in the retail and consumer products industry, please visit: www-935.ibm.com/industries/retail. Follow IBM Industries on Twitter at @IBMIndustries.

Back to top

rw-new-72-R-300wide

About RetailWire

RetailWire is retailing’s premier online discussion forum, serving the industry as a free resource for over 15 years with compelling content that goes well beyond conventional headline reporting. Each business morning, RetailWire’s editors post timely topics worthy of commentary by the RetailWire BrainTrust panel of industry experts and general readership. The results are virtual round tables of industry opinion and advice covering the most dynamic trends and issues affecting the retailing industry.

Contact:

Al McClain
almcclain@retailwire.com
(561) 627-4974

Back to top