Optimizely Personalization Presents: Adaptive Recommendations
Earlier this year Optimizely Personalization launched Adaptive Audiences — a feature that uses Natural Language Processing (NLP) to dramatically improve the way you can personalize your website. Today, Optimizely is introducing another big update to Personalization, Adaptive Recommendations — the only recommendations solution with experimentation at its core.
The Power of Personalization
Experimentation is all about learning what your users truly want and then delivering it to them. It is through experimentation that we learn people actually behave in manners quite different from one another. Companies like Netflix have built their entire business around this concept of personalizing the experience for every visitor. The movies I see in my account are different from yours, they are unique to my interests and behaviors. But it has been hard for other brands to fully adopt Personalization at the scale of a Netflix — part of this is because the technology and expertise hasn’t existed outside of these companies. Optimizely is seeking to change that.
Personalization is an area Optimizely has been focused on democratizing: how do you bring what’s great about A/B testing to the work of personalization? How do you know that showing a different experience to each visitor really impacts your business? Our answer to these questions is the newly updated Optimizely Personalization, the only experimentation-first personalization platform out there.
Our machine learning (ML) powered audience builder, Adaptive Audiences, has already proven to be a game-changer for companies like Crate & Barrel. And we know companies want to go beyond segment-based targeting, to offer what is considered the Holy Grail: always-on 1:1 personalization. So today, we are excited to announce something new: Adaptive Recommendations.
Introducing Adaptive Recommendations
Offering product or content recommendations has become commonplace in 2019. We’ve all seen recommendations in the retail and media experience online. But beyond those industries many have followed suit by adding recommendations to their sites; technology companies recommend what case studies to read, media companies recommend content, travel companies recommend trips and packages, and airlines present upgrades and mileage plan offers. However, few are testing their algorithms, their strategies, the placement, or the creative in the way that an Amazon or Netflix is today. That’s why Optimizely has focused on bringing our expertise in experimentation to the world of recommendations.
Optimizely’s Adaptive Recommendations allows you to test one algorithm against another. You can determine if co-browse, co-buy, most popular, recently viewed, or user-based performs the best in different areas of your site. You can also test the placement, creative, and the messaging of your recommendation module.
The Optimizely Difference
When it comes to recommendations we wanted to make it both easy-to-use while also being adaptable to your business needs. Here are some of the highlights:
The easiest, most scalable way to implement recommendations today. Implementing recommendations with Optimizely is done via our User Interface (UI), without having to do an expensive product feed of your catalog items. The UI also provides tools for QA, previewing your recommendations, and quickly building new recommendations modules without needing to rely on extensive development resources.
Customizable filters that adapt to your business needs. Want to quickly launch a new recommendations module that only shows sales items, new items from a seasonal release, or complementary categories? Filters from Optimizely makes it easy to adapt any of our pre-built algorithms to meet your specific business needs.
Test everything. Choose the algorithm that’s best for every moment; the product details page will likely be different from the checkout page. Now you can know with certainty what’s working and measure the impact of recommendations on your online revenue.
How To Get Started
We are really excited about what is made possible by Adaptive Recommendations. Unlike other recommendations offerings, Adaptive Recommendations is designed for testing, which means that not only can you improve your recommendations but gather valuable data as you go. Too many brands today aren’t rigorously testing their recommendations — but now they can. One Fortune 500 brand recently went from just implementing recommendations, to testing their impact on consumer behavior. Through testing they learned that recommendations contributed to $300M in annual revenue — 20% of their online sales that year. Imagine what they can do to this baseline number by continuously testing which algorithms are performing best across their site?
We’re also excited about what we will offering in the near future: a content-based algorithm optimized for text-heavy content. We will be using the same NLP technology that powers Adaptive Audiences, and you will be able to generate recommendations based on the complete content & context of your articles or product pages.
Optimizely Personalization is the natural extension of our experimentation capabilities and it’s all built on the same platform, the same client-side snippet, and the same enterprise scale. We’ve used ML to make the powerful, simple. The time consuming, fast. So you can spend time where it matters most: creating what your customers will love.
Adaptive Recommendations will be rolling out to all customers with access to Optimizely Personalization in the next two months. If you don’t have access to Optimizely Personalization today and would like to learn more, contact your account team to hear how Adaptive Recommendations could help your business. You can also watch a demo of our Personalization products.