Although the revealing of Google’s Recommendation AI has already been done during the company’s Cloud Next event in 2019, Google is now launching its beta version for its customers. A fully managed service — Google’s Recommendation AI — targeting retail businesses, has been designed to help in delivering personalised recommendation of products to customers at scale.
According to the blog post written by the product manager, Pallav Mehta, the move has been taken in sync with the ongoing shift of retail companies towards data-driven strategies and the increasing customer demand. To keep up their relevance in this competitive scenario, the retail companies now require to provide an ultimate personalised experience to customers. And one such way of enhancing the experience is by recommending them products matching their interest, preferences and need.
As a matter of fact, Google has been using the Recommendation AI across all its platforms like for their advertisements, search engine, and YouTube recommendation. It uses machines learning to understand the customer behaviour and changes the variables of pricing, offers and labelling accordingly. This claims to improve the click-through rates of the company and the revenue as a whole
Subscribe to our Newsletter
Join our editors every weekday evening as they steer you through the most significant news of the day, introduce you to fresh perspectives, and provide unexpected moments of joy
Your newsletter subscriptions are subject to AIM Privacy Policy and Terms and Conditions.
Also Read: Secret Behind Youtube’s Great Machine Learning Enabled Video Recommendations
Upgrading The Recommendation Solution With AI
Earning customers’ loyalty leading to better retention has been a critical concern for retailers amid this crisis. And thus, retail companies, rather than manually managing customers and curating recommendation models, should upgrade their process with artificial intelligence. The Recommendation AI focuses on each individual customer and stitches together with their buying patterns, which in turn helps them to serve with more personalised recommendations of products.
According to Mehta, the advanced system “excels in not only handling recommendations of long-tail products but also for cold-start users and items.” With the help of ‘context hungry’ deep learning models, developed by Google Brain and Research, leverages item and customers’ metadata to find insights across millions of items and continuously revise those insights in real-time. Thus, managing changing catalogues, evolving customer habits and shopping trends amid COVID.
Also, with Google’s Recommendation AI, retail companies no longer need expert programmers to write coding scripts to train the traditional recommendation models; instead, the platform provides a simplified model management experience for retailers. The API of this end-to-end personalised recommendation system based on deep learning ML models allows ingesting data of product catalogue and user information and requests for a recommendation based on the data.
To get started with Recommendation AI, the retail companies need to integrate their catalogue and user data, including the available tool and then import that data on to the platform. Once that’s done, the retailers can choose their model type and specify their objective for the same. This information would allow the model to get trained on the specific requirements. According to the company’ blog post, the first tuning and training of the model take about five days, before it can actually begin to recommend products for customers.
Such advancement in the recommendation model will help in scaling millions of items in a catalogue and will help companies recommend a relevant product from the same. Another critical aspect that has been secured by Google is the biases associated with popular items. The Recommendation AI has been designed to handle seasonality in a better way, even with lesser data.
Also Read: How Recommendation Systems Have Transformed Over Years
Implementation Of Recommendation AI
The Recommendation AI by Google competes with ‘Amazon Personalize,’ which is an ML service making it easy for developers to create customised recommendations for its customers, and Adobe’s AI-powered recommendation tool. However, according to many early adopters of this Google’s Recommendation AI stated the tremendous value achieved out of it.
Case in point — a multinational beauty product retailing company, Sephora stated that since implementation the company has witnessed a 50% increase in their click-through rate on their product page and a 2% increase in the overall conversion rate in different platforms.
In another example, one of the online retailers of consumer electronics in Switzerland, Digitec Galaxus, stated the importance of finding the required product amid the pandemic, which was seamless by Recommendation AI. The company experienced a 40% additional increase in their CTR compared to previous years.
With Thomas Kurian joining Google as the CEO, this initiative is believed to be a part of their ongoing project of directing its focus onto six key verticals — finance, healthcare, media, manufacturing, the public sector, and the apparent retail industry.
Also Read: Will Google Cloud Platform Be Taken Seriously Under Thomas Kurian’s Leadership?
Wrapping Up
Currently, the company has released a beta version of the Recommendation AI, and the pricing is based on its operations like training, tuning it per node per hour and amount of predictions requested. This new system allows retail companies to decide whether or not to keep a recommendation model active which provides them with better control over their finances.
One can try it for free here.
