Analytics Vidhya has long been at the forefront of imparting data science knowledge to its community. With the intent to make learning data science more engaging to the community, we began with our new initiative- “DataHour”.
DataHour is a series of webinars by top industry experts where they teach and democratize data science knowledge. On 28th April 2022, we were joined by Dr. Shanta Mohan for a DataHour session on “Artificial Intelligence in Retail”.
Dr. Shanta Mohan is a Mentor and Project Guide at the Integrated Innovation Institute of Carnegie Mellon University in Pittsburgh, Pennsylvania, USA. She co-founded Retail Solutions Inc. (RSi), a pioneering retail analytics company based in California, and ran its Global Product Development Team.
She started her career with a Bachelor of Engineering (Hons.) in Electronics and Communication Engineering from the College of Engineering, Guindy. She pursued a PhD. in Operations Management at Tepper School of Management, Carnegie Mellon University.
With an experience of close to 3 decades, Dr. Shantha worked in the Semiconductor industry. She worked in Software Engineering and led Kaveri Inc.’s professional services business as the CEO. She co-founded and led the global development team for Retail Solutions Inc. (RSi) for 13 years.
Are you excited to dive deeper into the world of AI? We got you covered. Let’s get started with the major highlights of this session: Artificial Intelligence in Retail.
The need to know technology and to get more technologically-advanced, altogether puts many challenges before us. And in this session, we’ll learn what kind of problems exist in artificial intelligence in retail, what are the opportunities, and then apply AI and machine learning to those problems. This datahour will focus only on how retail data is important to address many of the challenges that are present in retail.
Retail means selling in small quantities to the ultimate consumer and industry of such selling. So, it has got several stakeholders, they are the one who provides these products and services, and those who sell these services to the retailers, and then to the consumers of the product. By applying AI and machine learning in this case study, we can find the flows and patterns of all those people. And then we can use those to make the process more efficient, and make the customer happier.
The retail industry landscape focuses on three important dimensions, there can be many more but here we’ll focus mainly on:
Geographies: It deals with the whereabouts, or the particular location we are in. The location determines the challenges, and from this we can know what are the important criteria of that part/location.
Products: What kind of products in retail? The implications of this product, the challenges it encounters and what are the possible solutions one can think of.
Channels: How these products get to the customers and what its supply chain creates in terms of its unique challenges.
The Retail Industry has evolved a lot over the years depending on which part of the world you are in. In the West, you have all kinds of technological advances all in the way back in terms of beginning with the cash registers, credit cards, e-cash registers, information systems, w.w.w(world wide web), e-commerce, and most recently social media. These advances have made lots of differences in the western/developed worlds. And as early as 1962, when Wal-mart first opened BIG BOX retailing, things were never the same. Infact, now they don’t exist any longer though they do exist at a few places even today.
In the developing world this phenomenon is new and developing. For example: in countries like India “the kirana/corner shops” is the lifeline of retail.
Now e-commerce has taken up a big way in most of the countries but it’s still not that popular as traditional kirana shops. E-commerce creates its own challenges and its own opportunities. The location determines the kind of focus the retail should have on servicing the customers.
In the West, e.g. the US, the focus is on automating the stores, just to keep under control the cost of production as if the labour is expensive. Same principle led to the establishment of AmazonGo too.
Source: Dr. Shanta Mohan presentation
And in developing countries, the focus is on how to make products and services available to customers more efficiently for a better price and how do we get goods to very remote locations in a more reliable and speedy manner.
The focus is different, though more often than technologies look at what’s happening in the developed part of the world and wanna bring it to the other parts too. The developing countries also always focus on adapting new technology as if it adds to their progress.
The kind of products and services we get determine the kind of problems and challenges that the retailers and the suppliers encounter.
Images Source: Dr. Shanta Mohan presentation
The product characteristics determine how the supply chain is structured and what kind of issues that arise and how you get such products to the customers quickly such that their freshness or authenticity will be maintained.
The white goods such as fridge, A.C, etc, are the ones which we buy very few times during our lifetime compared to something like fresh fruits which is perishable and has a short life span. So fresh fruits need to be delivered quickly as compared to white goods.
Then apparel where we have consumers who have been attracted by these suppliers to buy a new fashion every season or more often. The kind of problems that arise in this sector is different because they are concerned more about servicing the customer with the latest. Same time there is a problem of wastage, eg, somebody may buy something over the internet and there is a size issue so return is initiated. So returns are a big problem in this industry.
The luxury good market here focuses on quality and customer experience. So, one does everything to meet customer expectations.
Lastly, virtual products like Netflix, financial instruments, etc, have their own requirements on how to meet customer demands and keep their engagement.
In all of these the aim is to be able to provide goods and services as efficiently as possible to maintain the customer base to keep them from going away to some other retailer. Therefore, the kind of solutions you can develop with AI is very important.
It’s mainly of three types:
The Brick & Mortar Retailers: Walmart, COSTCO, and today’s Amazon are examples of this channel. Although Amazon is an e-commerce site, over time it understood the need for the physical presence of mart in order to keep itself in front of the customers as well as address some of the challenges that arise due to e-commerce.
Source: Dr. Shanta Mohan presentation
ECommerce: Alibaba, Flipkart, etc, are examples of this channel. It has its own advantages and challenges. The trick in crafting the AIML solution is to understand the advantages and address the challenges.
Source: Dr. Shanta Mohan presentation
DTCs: Nestle, Cosper, etc are some of the examples of this channel. These cater to the customer directly. It benefits in knowing the customer more better and sum up the retailer who is providing the customers. The whole idea behind AIML of having the data and being able to discover about your customer is very appealing.
Source: Dr. Shanta Mohan presentation
Without data no transformations can be performed. So to do so you need to search data, collect it, make sure data collected is relevant and then apply analytics in order to understand what you can do with the insights found in the data.
Source: Dr. Shanta Mohan presentation
In retail, collaboration is the key in order to get benefit . It’s a collaboration because each entity is the main stakeholder in retail space and owns part of the data. For example, retailers have data such as inventory data, predicting data about customer buying patterns, etc. This data is called master data that has to be there for both retailers and suppliers to make sense of any insights that you get from the data. So, it’s not enough to have only transaction data of the sale, in order to understand it better, collaboration is important. It benefits all including the customer.
Source: Dr. Shanta Mohan presentation
The types of data we deal with in retail:
One principle applies when we talk about data i.e., garbage in, garbage out and if your data is not good enough to derive insights the data is useless.
In early days it used to be 3V’s of data:
Presently, this 3V’s of data has been converted into 10 V’s of data.
Source: Dr. Shanta Mohan presentation
Source: Dr. Shanta Mohan presentation
It helps in automating, making the product available in a frictionless experience where you can just walk-in to the store, either shop or not, and walk-out. There are no queues, no cash points, detectors and used technology will automatically perform their functions before you exit. For example: amazon go, 7/11, etc.
In order to survive in the business, stores need to innovate.
For example: ALDI an global retailers have innovated itself to survive in the market. If you wanna buy alcohol here they are monitoring your face and decide accordingly.
Another example is WATASALE announced in 2018 in India but today either they are vanquished or don’t exist.
Source: Dr. Shanta Mohan presentation
The challenges they have is AI enabled smart mirrors. For example, Lenskart uses this technology. This is done to get better faster sales and perfect customer satisfaction.
These use AI in different ways such as chatbots. Earlier chatbots were home-based but now with AI they are more enhanced and are robotic.
Earlier there used to be a Test-to-test method that has shifted to voice-to -text now such as Alexa, Siri, etc.
Recommendation systems : Recommendations showed to us on social media, this happens because of AIML. Through collaborative filtering your data such as which product you were searching for last, your buying pattern, etc will be traced and accordingly your feed will be managed that too involves you as well as similar users like you. Another is content based filtering that doesn’t involve other users but focuses solely only on your behavior.
Today’s better system combines both collaborative filtering and content based filtering.
Source: Dr. Shanta Mohan presentation
There are two supply chain:
First one and the traditional one involves supplier, distributor, retailer, delivery, last customer. In this the supplier supplies the material to the distributor, distributor to retailer, retailer to delivery or directly to customer.
Second comprises the supplier and the customer. The supplier itself provides delivery of the product and services to the customers.
The supply chain problems can be best monitored with the AIML help. The supplier can invoke AI to check whether the product is reaching the customer on the promised date or not, what will be the best way to use to do delivery fast, and how to manage returns. AIML use makes all this long process a little simpler and faster.
Source: Dr. Shanta Mohan presentation
As per the records, the E-commerce penetration in retail has doubled since 2020. It adds to a retailer’s carbon footprint. The increasing need for fast delivery has added up to an increased carbon footprint because in order to deliver at the earliest the club delivery concept is vanishing nowadays. Resulting in more shipping numbers and packaging issues. If retailers ship all the items in one go, then this could reduce carbon footprint to a great extent.
For example UPS, it’s a kind of delivery van which uses AIML. AI helps the driver in discovering the most efficient way of delivering the product.
Source: Dr. Shanta Mohan presentation
There are going to be so many new technologies to come:
I hope you enjoyed the session on Artificial Intelligence in retail and understood it very well. I hope by employing the use cases I provided you with a clear understanding of Artificial Intelligence in retail.
Lorem ipsum dolor sit amet, consectetur adipiscing elit,