How we are using AI to power content recommendations

In this article we walk you through how we are using the AI recommendation engine Recombee embedded in our headless CMS StoryBlok to drive content recommendations throughout our own website.

Whilst there has been a lot written about the impending age of Artificial Intelligence (AI) and how it will improve the way we live, advance healthcare and seemingly make ‘everything’ much better (or potentially cataclysmically worse depending on who you believe) - to date there has not been a lot of opportunities for businesses to actually use AI in any real or direct sense. For most of the last decade the ability to leverage AI technology has predominantly been limited to only the larger corporations who have the budgets and scale to accommodate the challenging learning curve associated with developing AI tools.

Happily though, that is starting to change with various different AI tools being produced that democratise access to the power of AI.

Introducing the Recombee Recommendation Engine

One of those tools is Recombee, an AI recommendation engine that provides businesses with the opportunity to leverage a combination of both content and behavioural analysis to automate recommendations. What’s more, whilst the “brains” of the AI are a bit of a black box, the tool does allow business inputs into its decision making enabling human tuning of the recommendations to meet business needs/imperatives. But what does all this mean in practice?

How does it work?

Let’s take the fairly simplistic example of our website. As with most websites, littered throughout our site we have little sections that suggest other pieces of content that you may like to read, as a way to encourage the visitor to continue their journey on our site and find out more about our business and our capabilities. Most commonly these appear at the end of our blog articles and our case studies.

The engagement with these suggested articles varies a lot from page to page. All of the suggestions are set up by our website content editors when they publish each article (or case study) and they usually try to find things that they think could be a logical next step for the reader. Unfortunately, unless we go through the laborious task of constantly updating them, the relevance of the recommendations can get dated quickly and on top of that the relevance of the recommendation is purely based on a "best guess" from the content editor.

Enter Recombee

Recombee is able to ingest key pieces of information about every page on our website and understand what they are about. This gives Recombee a contextual understanding of the content the user is viewing and it can then extrapolate out what other topics are related to that piece of content that would likely be relevant to the visitor. As a visitor looks at additional pages the AI engine is able to expand its understanding of the visitor's interests based on the content on the additional pages, and then factor this into its recommendations. But this is just the first third of the equation…

Recombee also ingests user interactions with the website and so it is able to layer over other users past behaviours in terms of what people click on vs what they don’t, and then add that as another layer to its decision making.

Finally, the Recombee AI tool allows the business user to fine tune the outputs by managing a bunch of different variables, such as putting a higher (or lower) weighting on a piece of content being shown or promoting certain pieces of content for certain content types, etc. It also comes with a strong data and analytics suite so you can analyse its performance and refine your approach as needed.

All in all this means we now have completely dynamic content recommendations throughout our site that will learn and adapt over time to improve the visitor experience and drive better results for our business.

How and where you can use a Recommendation Engine

As described above, a recommendation engine can be used incredibly effectively to recommend content. It is potentially even more valuable though when it comes to ecommerce applications, where you can empower Recombee to provide suggestions to shoppers that can drastically improve order volumes, drive repeat orders and enhance cart sizes.

It’s also worth noting that Recombee is not just limited to providing recommendation blocks on website pages -  it can also power search functionality, personalise email sends and even push notifications.

How much does it cost?

This might be the best bit, Recombee’s pricing starts from as little as USD $99 per month for 100,000 recommendation requests (which is a lot) and given the power of the tool it usually pays for itself very quickly.

Want to know more? We would love to talk with you about how you can leverage Recombee’s powerful AI recommendation engine for your business – just reach out for a chat.