Artificial intelligence (AI) and machine learning (ML) are among the top technology trends in the retail world. They are having a great impact on the industry, in particular in e-commerce companies that rely on online sales, where the use of some kind of AI technology is very common nowadays.
However, you do not have to be a big company or sell exclusively online to take advantage from the tremendous power of machine learning.
In this guide we will see how both online retailers and brick and mortar stores of any size can integrate machine learning technology to stay ahead of their competitors, by increasing sales and reducing costs.
From clothes to groceries to household items, the possibilities in the retail space are full of promise. The applications and use cases presented in this guide are a fraction of the feasible machine learning projects and serve as examples of what can be done today in the retail space. That being said, many companies have very unique needs that could be served with data.
You'll notice that every machine learning use case described has either a symbol, a symbol or both symbols at the same time.
These symbols indicate whether the use case corresponds to a machine learning project that:
primarily increases revenues.
primarily reduces costs through automation.
+ impacts both revenues & costs.
Customers enjoy browsing online catalogs because they present products in an attractive way and provide plenty of information about them. Although automatic digital catalog creators usually provide decent solutions, the use of custom machine learning technologies can significantly improve customers experience to increase engagement and impact conversion rates.
Many e-commerce and retail companies are leveraging the power of data and boosting sales by implementing recommender systems on their websites.
Companies using recommender systems focus on increasing sales as a result of very personalized offers and an enhanced customer experience. Recommendations typically speed up searches and make it easier for users to access content they’re interested in, and surprise them with offers they would have never searched for.
What is more, companies are able to gain and retain customers by sending out emails with links to new offers that meet the recipients’ interests, or suggestions of films and TV shows that suit their profiles.
The user starts to feel known and understood and is more likely to buy additional products or consume more content. By knowing and what a user wants and showing it right away, it less likely that it leaves the platform. This translates into a higher chance of purchase and a decrease in the threat of losing a customer to a competitor.
Providing that added value to users by including recommendations in systems and products is appealing. Furthermore, it allows companies to position ahead of their competitors and eventually increase their earnings.
Pricing is a critical predictor of profitability. Based on econometric science, a machine learning algorithm can consider key pricing variables into account, to define an automatic pricing strategy with real-time, dynamic prices.
The willingness to pay can be estimated from the customers behavior, for example considering the items they look at and purchase, or the time they spend on each web page.
The algorithm can constantly crawl the web, searching for information on the prices of your competitors for the same or similar products, considering hot deals and collecting enough information about the price history over the last number of days or weeks.
By including supply, seasonality, external events related to your business (e.g. a concert, a match, a festival), market demand, and offer, an automatic pricing system with machine learning can efficiently adjust and optimize prices.
A system that can learn most of what is happening in the market allows you to have better information than your competitors in order to make better decisions.
Read here the complete introduction to price optimization using machine learning.
Customers tend to search for visual content prior to making a purchase. However, in some cases they cannot easily find good keywords to describe what they want. The goal of visual search is to make much easier for consumers to find exactly what it is they’re looking for.
Instead of typing a query such as ‘cordless combo kit with soft case’, which will probably return a lot of general results, customers can upload an image to help narrow the search down to more specific items. With the huge and increasing amount of snapping and sharing images, machine learning algorithms can currently achieve amazing results.
This is one of the most trending use cases in online content. Leading companies like Microsoft, Google or eBay have presented in 2017 Bing Visual Search, Google Lens and Image Search.
Regarding the e-commerce market, Pinterest introduced Lens Your Look, a visual search engine that allows you to find outfit ideas inspired by your wardrobe. So if you are looking for new ways to wear your favorite jean or blazer, you can add a photo of it to your search to find outfit ideas that you eventually can buy.
In the solutions described above, machine learning models are trained with information about products, customers and purchases. Additional information such as reviews or feedback can be helpful.
|Recommendation Engines||Data from customers (purchases, searches, navigational data)
Data from products (images, descriptions, prices)
|Pricing Strategy||Data from customers (purchases, searches, navigational data)
Data from competitors (brands, websites, social media)
Data from products (descriptions, prices)
Data from external events (if available)
|Visual Search||Data from products (images, descriptions)|
If you do not have some of this data yet, you should consider that a lot of information can be obtained by crawling the web or by using specific services. Moreover, the 2016 Deloitte Millennial Survey states that half of shoppers and 58% of millennials would agree to share data if in return retailers would offer personalized services and offerings.
Therefore, it should be feasible to develop the mentioned solutions if you already have an online shop site or app. Regarding the volume of data needed, although more data tends to give better results, in some cases small datasets allow to have perfectly exploitable results.
The goal of a predicting customer behavior system is to estimate how buyers will behave in the future based on data of previous behaviors. These systems allow retailers to segment customers and perform personalized marketing actions that are more effective than general approaches. Moreover, taking actions based on predicted customer needs increases loyalty and retention.
A typical application is to predict purchases. For example, to know which customers are likely to make a purchase in the next 7 days. More complex predictions may have to do with important events in people’s lives. For example, to predict marriage or pregnancy, and then send custom offers.
Predicting the needs of consumers is a challenging task where machine learning algorithms are of great help.
The predictive models need basically consumer behavior data. That is, for example, purchasing history or buying trends, but it could also include social media activity and domain specific knowledge.
|Data from customers (purchases, searches, navigational data)|
How often your customers make transactions? Do they buy during sales time or before their birthday? How many items they usually buy? What are they currently buying? Of what topics are they talking in the social medias? All this kind of information is used by the models to predict future behavior.
And last but no least, retailers experience is very important to choose business specific criteria and fine tune the models.
Today, social media are more than networking platforms, and consumers are using it as a marketplace where they can buy products and services.
Monitoring social media on a large scale and obtaining valuable insights is possible thanks to the power of machine learning. Retailers can thereby get information about what is driving engagement, traffic and revenue.
By tracking and analyzing the flow of information, retailers can optimize the channel, target audience, content and timing of their social media posts and marketing campaigns.
Monitoring mentions of a retailer to get insights is a known application. But thanks to image recognition, retailers can now see how they are being portrayed through the images and videos shared daily. At the same time, this technology can be used to analyze and act upon content generated by their competitors.
This is a use case where a development from scratch is feasible, since retailers are usually aware of their most important competitors or the interests of their customers. Information about their social media accounts is helpful but not mandatory.
Chatbots interact with customers and simulate a human conversation, bringing them closer to the shopping that buyers get in a physical store.
They can provide added value at different levels. For example, chatbots can be used to encourage additional customer purchases, to personalize the customer experience, to improve searching capabilities over your catalog, or to handle a considerable part of your customer service.
The goal of a smart assistant is to emulate a human salesperson who might help you to find what you are looking for.
A good example of a smart assistant is Gifts When You Need (GWYN), by 1-800-Flowers.com, a floral and gourmet foods gift retailer. Through a natural conversation, a customer provides information about a gift recipient. Then the assistant tailors gift recommendations, based on the gifts purchased for similar recipients.
Smart assistants can have a tremendous impact. Within the two first months after its launching, 70% of 1-800-Flowers.com online orders were completed through GWYN.
When we know the terms to search, search engines are very good at finding data. However, sometimes we want a more sophisticated search, such as items of a certain shape or range of colors.
Since retail brands have usually large catalogues, they can be difficult to navigate, event when filters are available via a faceted search engine. An interactive chatbot can deal more easily with ambiguous requests and diffuse semantic, in the same way a human shop assistant would do.
Chatbots can be programmed with responses to frequently asked questions, relieving a customer service from recurring interactions. They can also include dynamic business oriented goals, such as answer questions about tracking and delivery, to improve the post-purchase shipping process.
A good chatbot can significantly reduce costs without losing quality, since it can detect when a complex query needs human intervention and then redirect the customer to a live agent.
A dataset of samples of conversations to be addressed by the bot is a very useful input, but it is not mandatory. For a contact center, a FAQ is usually enough input to develop a chatbot. For a smart assistant or an intelligent search system, more information about your catalogue will be probably necessary. But again, a chatbot can be developed in an incremental way, so it can handle bit by bit your products and services.
The more important thing is to know your business and be able to imagine the needs of your customers. Moreover, the are pretrained models for some kind of interactions that can be adapted to your use case by introducing specific business knowledge.
|Smart Assistant||Data from products (images, descriptions)
Samples of conversations (if available) or a set of use cases.
|Intelligent Search||Data from products (images, descriptions)|
Data from products (images, descriptions)
Optimizing inventory planning and predictive maintenance is a key issue and a very important logistic concern for retailers.
Machine learning algorithms can exploit purchase data to predict inventory needs in real time. Based on the day of the week, the season, nearby events, social media data and customer past behavior, these algorithms can provide a daily dashboard of suggested orders to a purchasing manager.
Brick and mortar retailers can take advantage of the impressive recent results on computer vision. The new approaches in the field could be used to generate real-time, accurate estimates of the products in a given store. With this information, a machine learning algorithm could notify store managers of unexpected patterns of inventory data that could be due to theft or an unusual increase in the demand for a product.
Another application is to use images to analyze the use of shelf space and identify sub-optimal configurations. A very good example of this technology is LoweBot, the Lowe’s autonomous retail service robot, that, besides helping customer to shop, constantly monitors inventory and gives real-time feedback to the store employees.
The inventory planning models need consumer behavior data. That is, for example, purchasing history or buying trends, but it could also include social media activity and domain specific knowledge.
Computer vision algorithms need images to process. They can come from the security cameras installed in the store or be taken by employees.
|Predicting Inventory Needs||Data from customers (purchases, searches, social media)
Data from external events (if available)
|Computer Vision||Store images and videos
Data from products (descriptions, images)
A good thing about physical stores is that the behavior and interaction of humans with products can generate valuable insights in ways that online retail cannot. Computer vision algorithms can recognize faces and people’s characteristics such as gender or range of age, generating precious exploitable data.
Where to put different items is a crucial matter for physical retailers, who always look for additional ways to understand the customer’s path to purchase.
Computer vision algorithms can track customers’ journey in stores to understand how they are interacting with it. These algorithms can detect the walking patterns and the direction of the gaze of the customers. Retailers can use this information to restructure store layouts or to measure the interest in their products. They can also discover locations that get a lot of traffic and visual attention.
Do elderly people shop more on weekdays? Do teenagers tend to cover only part of the store, for example the front part? Is the store more visited in winter? Variables such as age, day of the week or season could be used to generate insights that help to dynamically modify product placements and create efficient promotions.
Theft Prevention is a common problem in retail with a strong ROI, where machine learning technologies can go beyond the typical use of video cameras to detect shoplifters.
Facial recognition algorithms can be trained to spot known shoplifters when they enter the store. Walmart has tested this technology in 2015 as an anti-theft mechanism. In the same way computer vision can detect if someone picks an item, it could detect if someone hides an item in their backpack or jacket. Moreover, the same approach can be used to detect when checkout clerks skip scanning items, either inadvertently or on purpose.
A system based on machine learning can alert in real-time security personnel or managers and send them video excerpts so they can judge by themselves before confronting the individual in the store.
Brick and mortar retailers usually have no information about the items that customers pick up, glance at, and put back on the shelf. They do not have any information either about what customers look at next.
A computer vision algorithm can monitor shoppers facial and hand gestures to estimate how successful an item is. This kind of applications generate precious data about how many times an item is picked up from the shelf, put back on the self or in the shopping cart, or purchased.
If your store is equipped with security cameras having a certain image quality, then you already have all that is needed to start implementing the solutions previously mentioned.
Virtual or interactive mirrors bring shoppers a mixed experience between physical stores and digital interactions. This is a technology especially valuable for fashion retailers and makeup boutiques.
Instead of trying out clothes, virtual mirrors allow customers to navigate through virtual clothes models to find and choose the one they most like. Coupled to a recommendation system, virtual mirrors could suggest accessories and related clothes. Moreover, a “look book” could be created by combining outfits, so the shoppers can decide which clothes suit them best.
Cosmetic boutiques can also improve customers experience by letting them virtually try products. A specialized virtual mirror can also act as a beauty consultant, analyzing someone’s skin, looking for wrinkles, dark spots and clogged pores, and generating a report with concrete actions to be carried out.
This is a very hot market and there are several turnkey solutions in the market, such as HiMirror, a beauty and health consultant, or SenseMi, a mirror for fashion stores. Beyond the existing solutions, a retailer may need a custom development according to the needs of its company and characteristics of their products.
Some of the solutions proposed for online retailing can be integrated in mobile applications to get intelligent in-store assistants. These are among the most popular AI use cases in retail.
Such an assistant could tell customers the place where a product or a family of products are found in the store, help them to interactively find what’s best for their needs or propose them the best journey in the stores to pick up a list of products. Knowing their shopping list, the mobile application could suggest items that are probably missing from the list. In this scenario, a tablet or phone could be used as a pointing device that allow the customer to get instant information about items. Thank to speech recognition, interaction with the assistant feels more human.
An example of this kind of technology is Macy’s On Call, an application that helps shoppers to get information while they’re navigating the company’s stores. Another interesting use case is the Digital Tire Journey in-store web app proposed by Sears Automotive, which helps shoppers to find what they need among the great assortment of tires.
In-store assistants generate a double benefit. On the one hand, they provide real value to customers, which increases loyalty and retention. On the other hand, they enable retailers to collect a great amount of data that can be used as input by other machine learning solutions.
To develop an in-store assistant, information about products, inventory and customers is usually needed. In addition, an outline of possible interactions is necessary to begin development. Keep in mind that an assistant can be developed in an incremental way.
|Data from customers (purchases, interaction history)
Data from products (images, descriptions)
Data from inventory (store, products)
Samples of interactions (if available) or a set of use cases.
At Tryolabs we build custom data-driven solutions to improve company’s KPIs. This means that we partner up with retailers to conceptualize & build custom machine learning systems that either increase revenue or reduce costs.
Download here the free ebook with 10 real-world machine learning case studies that we have supported at Tryolabs.
Main contributors of this guide: Javier Couto, Martín Fagioli, Guillermina Umpiérrez and Esther Rietmann.