Content recommendations help keep people on your website, even after they’re finished reading the articles they came for.
According to data from Parse.ly, content recommendation systems generate a 3.2% uplift in page views. Savvy publishers have been able to get that percentage up even higher by using artificial intelligence (AI) technology to predict what visitors will want to read.
With the potential for a large payoff in the form of decreased bounce rates, local publishers have been eager to add content recommendation widgets to the bottom of most articles on their websites. Let’s learn more about how local publishers are using AI for personalized content recommendations.
What Are Content Recommendations?
Local publishers who want to keep readers on their websites for longer will often add widgets to the bottom of each story that suggest additional stories that readers may want to view.
The articles that populate in these widgets are usually the result of manual related content tagging or a third-party recommendation system that uses algorithms to generate the links that appear at the bottom of each article page. From a functional perspective, that means publishers are adding outward links, either manually or with CMS logic, with the vast majority of website visitors seeing the same stories below each article.
The way these stories have traditionally been selected is by popularity or by subject. For example, someone reading a story about a new restaurant might see recommendations for additional restaurant reviews at the end of the article. If a website’s content recommendations are based on popularity, then that same reader would see the most popular articles show up below the story he or she was reading, regardless of the topic.
Content recommendation widgets placed below articles on local news websites have a proven track record of success, but forward-thinking publishers are always looking for ways to improve, and that’s where AI comes into play.
How AI Improves Personalized Recommendations
Now that we understand how most content recommendation systems work, let’s look at how to make them better.
When publishers use popularity or topic to populate their content recommendation widgets, they run the risk of readers seeing the same content on multiple pages. If a reader didn’t click on an article the first time he saw the link, there’s an even smaller chance he’s going to click on it the second time.
With AI, algorithms can be programmed to recommend articles to readers based on the content they’ve already consumed, combined with any other data the publisher has been able to collect.
What does that mean in the real world? For starters, a reader who spends significantly more time in the sports section than the dining section is more likely to see sports story recommendations, regardless of which section of the website he’s on currently. Location and demographics can play a role here, too. Depending on how the content recommendations algorithm has been setup, publishers could suggest hyperlocal content that’s focused on the reader’s own neighborhood.
The more personalized the content recommendations, the more effective they become.
Content recommendations fueled by AI get more relevant over time. Within a matter of days or weeks, a recommendations algorithm should be able to determine which articles a reader is most likely to click on, so that readers can be exposed to even more of the content they might appreciate.
How Common Is AI in Content Recommendations?
The practice of using AI to improve content recommendations is growing within the local publishing community. According to a survey by Reuters, 59% of publishers say they’re “actively looking into” using AI to improve content recommendations on their websites and mobile apps.
Local publishers with WordPress websites have a number of content recommendation widgets to choose from.
A widget called Contextly gets readers to explore websites by identifying “evergreen” content—that is, articles that are always relevant—and the most popular articles, and then including those articles on a recommendations tab that publishers can add near the bottom of each page. For larger websites, Contextly also offers a personalized recommendations feature.
Other popular content recommendation widgets for local publishers with WordPress websites include Bibblio, which uses “smart AI” to keep readers on websites for longer, and AddThis’ WordPress Related Posts Plugin. Both plugins generate automated recommendations based on what’s most relevant to readers.
If you’d like help adding content recommendations to your own local news website, our team here at Web Publisher PRO would be happy to help.