Applying AI in The context of eCommerce
Artificial Intelligence is going mainstream and moving out from the realm of buzz to embedding itself into our lives — from self checkouts, airport security, email , marketing and to recommendations AI is everywhere.
A common question which I often get asked is “ How do you apply AI in organizations? “.
The term AI and machine learning gets used interchangeably although machine learning is only a subset of AI. The top players like Amazon, Google and Microsoft along with others are investing into AI related technologies with the intent on creating a business from AI based platforms and services.
As tools and infrastructure to perform machine learning (and deep learning) become available on demand as a service (mLaas) this becomes an ever important question to address. Just like with a startup — the fundamental question boils down to What problem are you trying to solve? Or what goal is your organisation trying to accomplish?
It really depends on your business and its context — what insights can you gain from data available to you both from within and outside your organisation which help you either 1) gain a competitive advantage 2) learn more about your customer 3) anticipate or predict your customer demands or 4) provide additional value to your customer so they further engage with additional offerings
Lets take the example of a retail house which sells luxury fashion items to customers worldwide through its group of stores and complements with a online e-commerce store. You may want to leverage the data you collect about the customers, sales and operations to derive insights from this data using variety of ways including machine learning.
Machine learning can generally do two types of broad outputs — regression and classification. On a high level, with regression we can predict a continuous quantity (how much will this item cost given its feature set) while with classification we can predict a discrete class label. (is this transaction fraudulent? or is this customer likely to purchase again?).
Lets look at some core areas where AI could help eCommerce -
- Recommendation systems to cross sell additional products
Machines can automatically recommend personalised products to your customers after analysing their past behaviour. We see amazon doing this wonderfully well and other e-commerce players have begun to adopt smart recommendations on their e-commerce sites. For quick time to market, experiment with recommendation as a service systems like recombee , nosto, bariliiance and others.
2. Personal shopping and customer service assistant as chatbots
Provide customers a conversational interface to ask questions, get recommendations, shop and even provide customer service using the power of a machine (chatbot). Research has shown that 80% of routine questions can be correctly answered and reduce 30% of your customer service costs. What you are providing is another interface which the millennial generation is very comfortable interacting with.
3. Assortment intelligence
Intelligent systems can present the right assortment of products and prices (from past customer patterns) to improve order size and also offer dynamic pricing.
4. Supply chain management
In a global context — the machine can analyse hundreds of variables and factors with present and historical data to predict demand for certain products or brands in geographies to help you with the management of supply chain and inventory. AI technology allows the organisation to get important information about factors driving demand and predictive analysis for what future demands are likely to come into the market
5. Personalised marketing strategies.
Once a machine understands customer behavior — it can tailor make campaigns. Once it analyses customer responses for a set of campaigns, it can become smarter with time and send intelligent tailored messages to maximise the desired response.
6. Predict customer behavior
Your machine learning algorithms if trained correctly can predict customer repeat buying patterns, their affinity to sign up for loyalty program or their inclination to purchase during a certain season or in response to a particular type of promotional campaign.
7. Detecting fraudulent transactions and fake reviews
Machine learning algorithms can learn the nature and characteristic of fraud transactions and help classify transactions as fraudulent with a high degree of accuracy once it is trained correctly. The same principle can be applied to customer reviews on your website.
8. Image classification
It is possible to provide intelligent image based interfaces for customers to point, click and purchase from composite images. Amazon provides another service from its mobile app where you point the camera to any physical item from your phone and it pulls up related and similar items allowing you to purchase them. Using image classification and recognition, it is now possible to engage customers to create a different buying experience.
Summary:
Machine learning is here to stay and given the right amount of training and data it has some interesting possibilities in the domain of classification and segmentation.
I have just touched the surface of what’s possible and may not have covered all possibilities — implementing all this would require data from your ecommerce systems fed into the machine learning framework / chatbots so that it can learn and perform some of these predictions “intelligently”.
Hopefully it has given you enough to begin thinking about AI use cases. My suggestion is to approach it based on your business priorities, one which would create maximum business value and aligned with the goals your organisation aims to achieve.
Altaf Rehmani is a Technology Innovator, helped various businesses with Digital transformation projects, Agile Evangelist and a champion of applying technology to enable business growth. He lives in Hong Kong and can be reached via email or twitter. Please leave your feedback and a clap if you have liked this article.
Other articles which may be of interest:
Managing High Performing teams
