Machine Learning and Artificial Intelligence in Retail and E-Commerce : A Study of its Applications and Implications

Comments · 1177 Views

Machine Learning and Artificial Intelligence are one of the most widely used technologies in the development of current businesses. Each and every business is aiming to automate more and more of its operations to reduce cost of operations as well as reducing the time taken while performing them manually. Today artificial intelligence has made its way into every domain and in this article I will be analyzing how machine learning and artificial intelligence has changed the retail and ecommerce industry and what the future looks like. I will also cover the different ways companies have been finding solutions to their problems using ML and AI and try to analyze their advantages as well as their disadvantages.

In the modern era businesses are in a race against each other to provide better services at a more affordable price and in order to achieve this, they are tending to explore new technological domains for their answers. One such industry is the retail and ecommerce industry which has a high amount of competition and brands are on a constant lookout for a way to excel in their services. 

As per the above data the retail ecommerce sales increased from 1336 billion USD to 4479 billion USD in just a period of 7 years. Most CEOs in the retail industry believe that the credit for this boost should go to the advent of latest tech trends in the industry. Amazon CEO, Jeff Bezos, believes AI and machine learning will improve every institution and organization in the future—including his own. In retail, artificial intelligence is being adopted rapidly - between 2016 and 2018 there was a 600% increase in adoption. Companies are using Machine Learning and Artificial Intelligence right from their background operations to customer delivery tasks. According to Accenture, 58% of consumers are more likely to buy from a retailer that offers a personalized experience (which is one of the major use cases of ML and AI in the retail industry). In this article I will talk about some of these ways and how they are turning the tide of the industry with these projects.

Why is AI and ML needed in Retail and Ecommerce?
Let us try to understand why firms are opting to move to AI and ML solutions rather than sticking to their traditional sales plans:    

  • Lack of Personalization: Often companies try to make products that have been customized as per the demands of a particular customer, but it is very difficult for huge companies to cater to the requirements of each and every customer. Hence they need a process which can individually understand the needs of a particular customer and then customize products for them. This is where Machine Learning comes into action. With the help of chatbots, companies can suggest the customers what they are looking for and some companies are also able to customize their products as per the requirements of an individual.
  • Loss of Revenue: With every incomplete sale due to cart abandonment, e-commerce firms tend to lose almost 70 percent in revenue resulting in millions of dollars each day.For example, AI-based tools benefit the e-commerce companies by way of automating data, stock and inventory analysis that facilitates better forecasting of sales. AI-based chatbots can help allure customers with incomplete sales and abandoned carts, by showing them offers and discounts in order to induce a purchase. 
  • Unlock full potential and increase profitability: One of the biggest hurdles as per some companies is the time wasted on carrying out redundant processes manually for a long period of time. Most firms feel that they are able to be much more productive and optimal once their processes are automated and this is where Machine Learning and Artificial Intelligence help them. Not only does it reduce the time of their processes but also increases their profitability. For example, a company like Uber Eats is able to cater to more customers once a model is able to guide them on the shortest route possible and also allows to serve multiple customers who are located in the same locality at the same time.

Now that we know why the companies tend to incorporate Machine Learning and Artificial Intelligence into their operations, let us discuss some of the big firms that have done so, what steps have they taken and how they are setting an example for others to follow.

Looking into a few interesting use cases:
Most of the companies have stepped into this field and have been working with the solutions in some way or the other. However, there are a few major players that have been setting benchmarks with their solutions and have been a role model for others. 

We can see in the above image how companies generally use AI and ML in their operations. Now let us discuss some of these companies and the innovations they have come up with.

  • Amazon: Amazon is a brand that is well renowned for its fearlessness to step into new fields and for using new technologies. Amazon has been using Machine Learning and Artificial Intelligence since a long time now. Ranging from features as simple as product recommendation to as complex a product as DeepLens, ML and AI have had a huge hand in the success of Amazon. Amazon’s proud of its virtual assistant Alexa which thrives on technologies like these. The latest grocery store Amazon Go is a marvel of its kind which takes automation to a whole new level in this industry. 
  • Walmart: Walmart is another multi-billion dollar retail company that has its hands full with innovation, all thanks to Machine Learning and Artificial Intelligence. Walmart is said to have filed over 1,500 patents related to smart shopping carts that have features ranging from measuring shoppers’ heart rates to a temperature - controlled vehicle. It recently also has shown a lot of interest in a self-driving delivery project by announcing a partnership with Ford. Walmart has also been testing autonomous pick-ups with Waymo, a Google project that has since become a stand-alone subsidiary of Alphabet Inc. Walmart also has been using Machine Learning since a long time now to optimize its delivery routes and provide faster checkouts on the basis of the user's browser history.
  • Uber Eats: Uber has been known for heavily implementing Machine Learning in their eats business and in the newly acquired grocery delivery business too. It has been using ML to estimate trip times and forecasting the demand for their services. It has also been using machine learning to speed up customer support resolutions. It is estimated to reduce the time taken to resolve the tickets by 15%. Uber also recently launched its advanced ML platform, Michelangelo.
  • LOWE’s: In 2016, Lowe’s introduced LoweBot, an autonomous retail service robot designed by Fellow. LoweBot was able to find products in multiple languages and help customers effectively navigate the store. As LoweBot helps customers with simple questions, it enables employees to spend more time offering their expertise and specialty knowledge to customers. LoweBot also assisted with inventory monitoring in real-time, which helped detect patterns that might guide future business decisions.
  • ASOS: ASOS is an online fashion outlet and has been a pioneer in shopping the applications of Machine Learning in the fashion retail industry. It has been using image recognition for tagging its products with different labels. Also, it has been using different features to train a model to give each customer a CLTV (Customer Lifetime Value) score depending on which it labels its customers as regular and loyal or new customers and roll out personalized offers on this basis.
  • Caper’s Smart AI Cart: Caper is a startup aiming at making smart shopping carts which will make autonomous stores more feasible than setting up cameras on aisles and ceilings like Amazon Go. The smart cart will be able to use image recognition to detect the items the customer puts in the cart and will weigh them accordingly and once the customer is done shopping it will allow the customer to pay using their cards in the system attached to the cart itself and the receipt will be sent to them via email.
  • North Face: The North Face is a 48-year-old, U.S.-based outdoor apparel and products retailer. North Face recently teamed up with IBM’s Watson and software builder Fluid to create a smart chatbot that gives their shoppers a personalized experience. It asks and tries to understand the customer’s needs and answer their questions which in turn gives the company a more loyal and happier customer base.
  • Wayfair: Recently wayfair, the famous online furniture store, started using Augmented Reality in their mobile application which would enable shoppers to visualize a piece of furniture in their homes even before buying it. The company believes this would save a lot of money spent by the company otherwise in returns and exchanges, since customers will be sure before buying whether they like the product in their personal space or not.
  • Aspect by Digital Bridge: Following on similar steps like Wayfair, Digital Bridge also created an AI model which allows users to customize their bathrooms as per their wish with the help of AR and image recognition. Customers are able to design their bathroom themselves and then share them with a professional to work on it, thus smoothing the process and speeding up things.

Barriers in Machine Learning and Artificial Intelligence
Companies are investing huge amounts of money in these technologies and reaping the benefits they provide. However this is not the case with all organizations, there are huge barriers that still exist in implementing these solutions and we will discuss and analyse some of them in this section.

  • Lack of Vision: One of the major reasons why companies are unable to step up their investment in this space is because they lack an implementation plan. Companies often need a clear long-term as well as a short-term plan which many of the companies today are unable to make. Companies often continue operating the traditional way rather than transitioning to latest tech trends and hence tend to fade away.
  • Level of confidence: One major drawback with such solutions is earning the level of confidence on the services provided by them. For example, it is very easy for an image classifier to misclassify a 1 pound bag of sugar to a 1 pound bag of salt and hence charge wrongfully to the customers. Such a solution would not only decrease the reputation of the brand among customers but would also force the companies to remove these solutions which inturn would lead to a huge loss of capital spent on buying them in the first place.  
  • Quality of Data: The most important component of any Machine Learning model is the data fed to it. It is really hard to get well labelled clean data and even if it is found, getting the required information from it is another challenge because of the quantity of data. As per IBM, 80% of the data that we get is unstructured and less than 1% of the overall data is being analyzed today for training models. So, certainly there is huge scope for improvement in this area but the challenge is to get good quality of data which is actually useful to the companies.
  • Lack of skilled labor: Most retail organizations often lack the required workforce for creating new technologies and hence thinking out implementing one comes with a challenge of hiring skilled labor. An organization needs to hire a whole range of employees right from data scientists to machine learning experts and program managers who have experience in this field. This comes at a huge cost for these organizations and companies often tend to refrain from them.
  • Lack of capital required: As mentioned above, companies need to spend a lot when planning to step into new innovation. Not only do they need to hire the right people for the job but also invest in the infrastructure demands of these solutions which is many times more than any other cost. This acts as a big roadblock to small and mid level enterprises which are new to the industry.

How to overcome these barriers?
Even though some of these barriers are huge and often firms tend to retreat their steps facing them, some of them do have a way around. Companies need to research well on these solutions and then come up with a plan on how to go about with them. I will discuss my personal thoughts on how companies can bypass these barriers and enter profitability.

  • Identify narrow use cases, well-supported with data: Firms tend to mostly get stuck with these solutions in the process of setting up a well defined use case. They often end up defining a broad use case which needs a lot of research and a huge investment in order to cover it. These organizations need to stick to a narrower use case and dig deep only when they have substantial amounts of data to back them, since lack of data is one of the biggest and most challenging hurdles in defining an AI model.
  • Choose open-source:  Firms entering into AI and ML often fear the level of investment needed in order to hire a skilled workforce for setting up a project of such huge scale. In order to avoid this, initially companies can go for open source softwares and companies providing AI solutions, work upon them to find out the feasibility of their plans. They can get quick results on smaller projects and be able to assess if they want to move forward or step back and rethink.

Even though there are several barriers to such implementations but not all are doomed. Implementing such solutions do have huge advantages ranging from making better decisions to optimizing business operations, automating processes earlier done manually and most importantly, improving profitability by forecasting and predicting sales.

The above figure (CREDITS: KUNGFU.AI) mentions some of the advantages companies felt after switching to ML and AI based solutions. In this section we shall discuss some of them in detail.


  • Optimizing Processes: Companies like Uber are able to optimize their processes which include finding the nearby restaurants and deals, optimizing customer tickets resolution time etc. On the other hand companies like Amazon and Walmart are able to optimize their backend operations like delivery times, warehouse management etc.
  • Personalization: Companies are able to provide customers with a more personalized shipping experience with the advent of machine learning. This is due to the fact that companies are able to better understand the needs of their customers and forecast the sales and provide customers with better personalized options. As per a report by Boston Consulting Group (BCG), retail stores that have opted for personalization have reported a 6 - 10% growth in sales.
  • Explore new markets: With their current processes optimized and automated, companies are able to step into newer markets and unlock their potential more than ever. For example, Uber recently stepped into the grocery delivery business and Amazon was able to go to Amazon Go stores rather than being an online store. Companies are able to step into new markets because they are able to forecast the demand and take the right steps accordingly, thus reducing their chances of losses.
  • Economic benefits: As mentioned above with the help of Machine Learning and Artificial Intelligence companies are able to save a huge amount of capital by forecasting sales, automating processes, optimizing operations and saving on labor needed. Apart from these savings, the companies are increasing profitability as well due to increased sales and sales from entering new markets. According to a retail executives survey by Capgemini at the AI in Retail Conference, the application of the technology in retail could save up to $340 billion each year for the industry by 2020. An estimated 80% of these savings will come from AI’s improvement of supply chain management and returns.

As clearly visible from the above data, almost all of the companies are able to show moderate to substantial benefits due to incorporation of AI solutions in their industry.

Talking about the barriers of implementing AI solutions is one aspect. However, once implemented these solutions often tend to pose some challenges due to which some major companies are refraining from switching to them. In this section we shall talk about some of the negative outcomes of these projects and try to understand them better.


  • Customer Privacy: In any model data is of utmost importance but generally these data points are the customers and their information. Such models often tend to collect some sensitive information about the customers which often does not go very well with the customers, since not all of them are happy sharing their information. Also keeping account of such sensitive information makes it very prone to security attacks which have been happening since many years and millions of these data points have been compromised.
  • Loss of Jobs: Another major disadvantage of bringing in this technology is the replacement of regular manual labor. This is due to the fact that companies often need much less workforce once their operations are optimized and autonomous bots are able to achieve better efficiency in certain cases.
  • Lack of human touch/interaction: Most people feel that shopping is best done when a human is able to understand their needs and are able to talk to a human. They feel that this is a barrier which comes in when talking to a bot since it demands very specific replies and this often leads to frustrated and unhappy customers.

What does the future look like?
The big question which now arises is how the future looks like in this industry given the current scenario. Are these solutions going to have a long term impact or will the profitability reach a peak? What more options are available for the companies to explore? What more solutions are companies trying to incorporate in the recent future? Will implementing these solutions mean loss of jobs? These are some of the questions that arise in everyone’s mind, so let us try to understand what actually the future holds for the retail industry.
First of all let us talk about what companies are working on for the future and if it is actually possible to practically implement them. Recently Walmart pointed towards using image recognition in order to classify the customers as happy or sad on the basis of their facial expression and body language. Moreover according to a patent application Walmart filed, it seems like its next step is integrating IoT tags to products in order to monitor product usage, auto replace products as necessary and monitor expiration dates or product recalls. These sensors would rely on a variety of technology such as Bluetooth, barcodes, radio frequencies and RFID tags that would provide Walmart with an incredible amount of data including the time of day products are used to where the products are kept in the house.Such steps would help Walmart gather a lot of beneficial data and also achieve customer satisfaction, hence pushing to establish itself as an even better brand. 

Many other companies like Uber, Amazon have also been investing in a lot of projects like delivery by autonomous vehicles and in projects like Uber Elevate which is shared air transportation between suburbs and cities, and ultimately within cities. I personally believe that in the near future we might see these projects turn into reality and would certainly be highly profitable for the companies. The global market for Artificial Intelligence Retail Business is expected to grow to over $5 million by 2022.
While these projects have been benefiting the people in one way, it is also a source of major concern for the society in a completely different sphere. People are scared that automating stuff and using robots will ultimately prove to be a job killer and most companies also support this belief. However companies like Amazon believe in training the workforce to achieve a higher skill set so that they are able to relocate them to a different job rather than removing them from the ecosystem.

I personally feel that the retail and ecommerce sector has made a giant leap in a very short period of time with the help of Machine Learning and Artificial Intelligence. On the other hand, it would not be wrong to say that we have barely scratched the surface and there is a huge potential remaining to be unlocked. We can achieve some truly remarkable milestones with the help of AI, ML and related fields. While these solutions have some challenges and drawbacks, I believe that the companies should not shy away from exploring the benefits. Eventually it is the responsibility of a company to take decisions responsibly and maintain a balance between their operations. As more and more firms are stepping into this field, the smart decision would be to aim for collaborative growth so that not only they, as a firm, benefit from the rewards but also we as a society are able to cherish them without any drawbacks.