Content Recommendations in Practice — How Our Customers Do It
Andy Thompson from digital agency Luminary talks about his first experience with Smart Content Recommendations in Kentico Cloud.
In the previous blog post about Smart Content Recommendations, we introduced the new feature and talked about how you can replace a tedious process of manual personalization with a simple solution to provide your customers with relevant content. In this article, we will reveal a use case of one of our trusted partners, Luminary so that you can see it in practice.
Personalization Made Easy
I reached out to Andy Thompson, CTO of Luminary, and asked him few questions:
Could you tell us more about your motivation to use the recommender?
Our new website had been designed with "read more", "you may also be interested in", "there's more where X came from" style sections at the bottom of most of our key content types/templates, such as blogs, case studies, and people profiles. We had designed a complex, custom algorithm to select suggested content based on "featured" flags, high level technical/marketing taxonomy, and using cookies to track which articles a specific user had read, and it was already getting extremely complex. At this exact moment, I was at Kentico's office in Brno and spoke with the team who were developing a machine-learning based recommendations engine prototype, which would do exactly what I was already trying to build in my custom code, but also take it further to make recommendations based on other visitors' activity. I nearly dislocated my shoulder throwing my hand in the air to help test it.
Where/how do you use it on the website?
At the moment we're using it for "read more" style-related content on blog posts and case studies, as well as similar/interesting staff profiles at the bottom of each staff profile detail page. We're also using it to generate some recommended blog posts and case studies on our actual home page.
What do you see as the biggest benefit?
By far the biggest benefit is that it keeps our related content up to date, on every single page of the site that it's used, automatically. Related content is always planned, and is always underestimated in projects. The amount of relationships you need to maintain (and therefore time and effort you need to spend) expands exponentially as you add more content, and in reality, people very rarely go back and update relationships for older content when new content comes out, it tends to be one way. Now we have better-related content, and it is completely automatic. Obviously, there are other flow-on benefits from having good related content too, such as better SEO, more pages per visit and so on.
How did it change your way of working compared to a traditional personalization?
It changed our way of working - we no longer have to do any work! Honestly, the biggest change is that now it actually happens. And not just sometimes, but every time, because it's completely automatic. Personalization, and in particular content recommendations for a content-heavy site, requires a lot of work. It's often planned but underestimated, and it is often the thing that doesn't quite make it if/when you're running out of time.
Try It for Yourself
If personalizing your content sounds like something you'd like to get started with, contact us and we'll help you to leverage Smart Content Recommendations across your existing or new projects.