Future of AI in e-Commerce: Transforming data with ML and Computer Vision


By Vineet Grover, Sr. Vice President, Analytics & AI Solutions, Paxcom

As businesses continue to innovate, strive against competition, thrive, fall and grow again – one thing they all need is Agility. Agility in decision making, backed up by efficient data analysis, leading toconviction in actions,repeatedly – untiringly. 

E-commerce is poised to grow to a $70 trillion market by 2028 according to (IMARC, n.d.). A lot of agile decision making would be needed for this to happen, and artificial intelligence (AI) would need to contribute significantly to this kind growth.We have already seen the role of data analyticsin areas such as customer segmentation, recommendation systems, fraud detection, customer support, sentiment analysis, chatbots, lending a great helping hand to business, especially in digital businesses and e-Commerce. 

With the latest advancements in generative AI like ChatGPT- there surely is a concern on the impact of AI on our lives if it is uncontrolled. But the way AI has helped, and can help human and business endeavours,in a controlled manner, with privacy or security concerns handled, is an aspect which would continue to grow and thrive.

In this article we would explore some more areas where AI would help businesses. Some of these concepts would hold true for both online as well as offline businesses. 

Marketing Mix Modelling 

Marketing Mix modelling refers to the process of finding potential value of all marketing inputs and other investments to find the best drivers for long term growth. Predictors like Ad-spend on different channels, mediawith all associated effects & lags, pricing impacts, product assortments, industry trends, competition, and other exogenous factors may be modelled for performance driver analysis.Sift through different kinds of models and find the best model for your business. Although it starts like an attribution model (what and why), it then grows into apredictive, prescriptive analytics (future and action), where it can truly shine. Inputs can be tweaked to understand the potential impact on outputs for better decision making. This can be a powerful tool for CMOs, e-Commerce and Sales heads of an organization to plan and map their growth opportunities.

Dynamic Price Optimization

A fundamental problem affecting all online businesses – How to dynamically set the prices for success, as things change quickly in online world.One would typically start with reviewingprice effects of mark-up and markdown on the perceived value. It suddenly explodes out to include all forces like demand elasticity, competitive benchmarking, category dynamics and your own unit economies (inventory levels, run-rate) to arrive at the final price. All of this needs a lot of time, and by the time you arrive at an answer manually, market changes. A well-designed AI model would do all of this for you. It would also keep updating its information and it can, potentially, learn from its mistakes as well.

Computer Vision 

It is a branch of AI to extract meaningful information from images, videos etc. While a lot has been discussedin areas of face recognition, graphical capabilities, some of the following use cases in e-Commerce can greatly benefit from computer vision, image processing.

Advertisement and Banner Tracking

As a Business you spend a lot of money for your Ads, but you do not always know whether the impressions for which you paid for were truly generated or not. Advertisement and Banner tracking solutions can track online creatives and keep you less anxious about those.

Offer Extraction 

As a customer you want to know the best offers on your favourite products. If someone can curate that list for you – what could be better than that? Usually the offers are in the form of Ad creatives – images and videos. Computer Vision models can extract information including brand names and offers. This information combined with the consumer interests can go a long way increasing the sales for businesses.

Visual search

AI powered search can help customers search items of their interest by submitting pictures of the product. Image similarity algorithms can do the job for customers in the background and lead them to the appropriate products.

Ad Creatives and Content Generation

AI is now at a stage where it can predict what sort of content and creatives would work for a target market. Not only that, it would even generate the needful for you. This is one aspect which every thinking business would try to take advantage of. 


Please enter your comment!
Please enter your name here