With the continuing shift to digital, especially in the retail industry, ensuring a highly personalized shopping experience for online customers is crucial for establishing customer loyalty. In particular, product recommendations are an effective way to personalize the customer experience as they help customers discover products that match their tastes and preferences.
Google has spent years delivering high-quality recommendations across our flagship products like YouTube and Google Search. Recommendations AI draws on that rich experience to give organizations a way to deliver highly personalized product recommendations to their customers at scale. Today, we are pleased to announce that Recommendations AI is now publicly available to all customers in beta.
Upgrade your recommendation solution
Instead of manually curating rules or managing cumbersome recommendation models in-house, you can upgrade your personalization strategy by replacing or complementing your existing solution with Recommendations AI.
By putting a greater emphasis on each individual customer rather than on an item, Recommendations AI is able to piece together the history of a customer’s shopping journey and serve them with personalized product recommendations. Recommendations AI also excels at handling recommendations in scenarios with long-tail products and cold-start users and items. Its “context hungry” deep learning models use item and user metadata to draw insights across millions of items at scale and constantly iterate on those insights in real-time in a way that is impossible for manually curated rules to keep up with.
Recommendations AI also delivers a simplified model management experience in a scalable managed service with an intuitive UI. This means your team no longer needs to spend months writing thousands of lines of code to train custom recommendation models while struggling to keep up with the state-of-the-art.
Key updates to Recommendations AI
You can now get started with Recommendations AI with just a few clicks in the console. Once you create a Google Cloud project, you can integrate and backfill your catalog and user events data with the tools you already use, including Merchant Center, Google Tag Manager, Google Analytics 360, Cloud Storage, and BigQuery.
Once the data import is complete, you can choose the model type, specify your optimization objective, and begin training your model. The initial model training and tuning takes just two-to-five days, then you can begin serving recommendations to your customers. To ensure that your setup is working like you want it to, you can preview the model’s recommendations before serving them to customers.
In addition to making it easier to get started, we’ve also been collaborating with the Google Brain and Research teams to push the boundary of what’s possible for recommendation systems. As a result, our models can scale to support massive catalogs of tens of millions of items and ensure that your customers have the opportunity to discover the entire breadth of your catalog. Recommendations AI is also capable of correcting for bias with extremely popular or on-sale items, and can better handle seasonality or items with sparse data. Our model training infrastructure allows us to re-train your models daily to draw insights from changing catalogs, user behavior, or shopping trends and incorporate them into the recommendations being served.
How customers are using Recommendations AI
Many retailers from around the globe have realized tremendous value from Recommendations AI.
Sephora, a multinational omni-channel retailer for beauty and personal-care goods with thousands of stores globally, is using product recommendations to personalize their customers’ e-commerce experience.
“We wanted to deliver the same highly personalized shopping experience to our clients on our digital platforms that they receive in our physical stores,” says Jaclyn Luft, Manager, Site Personalization & Testing at Sephora. “We started working with Google Cloud to explore how we could leverage its innovative machine learning technology to provide enhanced personalization to our online customers through product recommendations.”
“Since implementing Recommendations AI we’ve seen impressive results with a 50% increase in CTR on our product pages and a nearly 2% increase in overall conversion rate on our homepage relative to our previous ML-driven recommendations,” Luft continues. “We are now evaluating how we can continue to test, iterate, and expand the application of Recommendations AI to power recommendations on other areas of our ecosystem, such as within the checkout flow and in our emails.”
Hanes Australasia—home to many iconic Australian apparel and lifestyle brands—is another customer that’s powering personalization with Recommendations AI.
“Recommendations AI delivers extremely good data execution and shows how Google Cloud can turn data into real commercial value,” says Peter Luu, Online Analytics Manager at Hanes Australasia. “When we A/B tested the recommendations from Recommendations AI against our previous manual system, we identified a double-digit uplift in revenue per session.”
Luu also added, “the product is extremely easy to use—Google Cloud has provided the expertise, functionality, and performance, so we do not need to be machine learning experts to make the most of it.”
Digitec Galaxus, the largest online retailer in Switzerland that offers a wide range of products to its customers from electronics to clothes, uses Recommendations AI to help their customers find products they are looking for.
“With Recommendations AI, we are able to provide personalized product recommendations to our customers at scale throughout our website. Recommendations AI is also a great reference to test and challenge our in-house recommendations algorithms against.”
“During the pandemic, finding the product you need is more important than ever,” Sager explains. “In the past few months, we’ve noticed a strong increase in the usage of recommendations in general, with Recommendations AI performing with up to a 40% additional increase in CTR compared to the previous period. Customer needs evolved as the pandemic continued, and Recommendations AI adapted well to the changes and allowed us to keep up with our customers and their preferences.