by Amit Jnagal, on December 15, 2016 11:30:00 AM PST
ARTIFICIAL INTELLIGENCE IN eCOMMERCE PART 1
The gamut of Artificial Intelligence technologies such as image recognition, natural language processing, behavior-based learning, etc. can tremendously benefit as well as change how traditional merchant operations for eCommerce websites are performed. There are many areas where AI has the ability to positively impact merchandising. My objective in this series of posts is to cover as many areas as I can. However, for this post (PART 1), I will largely focus on how AI can benefit merchants to use enriched metadata such as product tags, descriptions, features, etc in their product catalog.
eCommerce merchants or anyone who manages a product catalog on an eCommerce website understands how critical it is to maintain and display relevant product listings and high-quality shopper-friendly product content. Maintaining fresh product listings, high-resolution product images, accurate product descriptions, competitive product features, etc. across thousands of products is where merchant teams spend most of their time and money.
Traditionally merchants have relied on manual quality checks, semi-automated methods, pre-configured rules, inflexible relevance algorithms, etc. These methods help them to ensure relevant product listings, accurate product descriptions, and comprehensive product attributes. Manual methods work, however they are expensive and time-consuming. In most cases, their application limits itself to best-sellers and top categories. Fully automated methods may address a few challenges. But these are also limited due to the rigidness of rules and their inability to keep pace with fast-changing data. Above all these systems do not understand product domains, do not self-learn, and mostly rely on inputs from the merchants to make decisions and take actions.
Creating, maintaining, and ensuring rich product content is a complex business. “You don’t know what you don’t know”. In this case, what it means is that proactively figuring out if the product content is accurate and of high quality is a tricky task. Merchants generally deal with the manufacturer’s content. They do not act until there is a question, a shopper makes a negative comment, or the conversion and engagement rate for a product is abysmally low. These steps are reactive in nature and generally, it is an operational nightmare for even a medium-sized eCommerce website. Traditional tools even lack the ability to integrate actions with real-time insights harnessed from vast amounts of user data, and social product content.
That’s where AI capabilities such as image feature extraction, natural language processing, sentiment analysis come into play and fare rather well. It addresses all the day-to-day challenges faced by eCommerce merchants.
Imagine a system that understands images (just like a merchant would) and enumerates all visual features for that product. Or, a system that evaluates thousands of user-generated product images and predicts pairing options or usage scenarios for a product. Or, one which can weed out low-quality low performing images.
Embedded image recognition and feature detection in merchandising tools can generate powerful insights and present information faster. It does not need any manual evaluation of the image content. We train the Image recognition models to automate feature extraction and translate them into product tags and features. It ensures that it translates every important aspect of a product image for search and SEO. It's like modeling a merchant's visual analysis in a machine model.
The potential for artificial intelligence in eCommerce merchandising operations is unlimited. I have covered but a tiny fraction of the possibilities. I will continue to share more in upcoming parts of this blog series. However, for now, merchandisers looking for “what’s next” should take the leap in 2017. They should initiate planning and implementing systems with AI capabilities at their core. This will help them to make their websites outshine those of competitors.
