Skip to main content

5 ways AI personalization can increase media revenues

September 26, 2019
Abdel El Ouazzani Kameleoon
Abdel El Ouazzani
Abdel El Ouazzani is an AI consultant at Kameleoon, and works with our clients on projects involving AI, machine learning and predictive algorithms. On the blog he shares his expertise and explains the value of AI personalization for specific industries.

1 Can we transform online media with predictive targeting?

Circulation figures for UK national newspapers dropped by over half between 2010 and 2018, and continue to fall. Many B2B and B2C magazines have closed their doors, while local and regional publications have cut titles and jobs thanks to smaller readerships and a drop in advertising revenues.

The reasons for this are clear. The move online has seen the market value of content decrease - essentially consumers are not prepared to pay the same amount for digital articles as for physical newspapers and magazines. This is due to the commoditization of information that the internet enables - essentially readers can find the news and articles they want at the click of a mouse (or the tap of a smartphone screen) without having to pay.

This has a knock-on effect on advertising revenues. Essentially publishers have swapped newsprint pounds for digital pennies, with larger revenues moving to the digital middlemen that control online advertising. In the US 70% of digital ad dollars now go to Google, Facebook and Amazon, according to eMarketer.

However it is not all doom and gloom for publishers. While technology may have initially been seen by the media sector as a threat to be controlled, or a challenge to be faced, it can also serve as an ally, helping newspapers and magazines find new growth areas.

In this article, I’m going to present 5 use cases that enable the media to increase and sustain revenue using AI-based predictive personalization.

What is AI-driven personalization? AI-driven personalization, or predictive personalization, essentially delivers tailored content to individuals, based on their preferences and behavior. Kameleoon’s AI platform can adapt messages, offers and content, in real-time, helping to both improve the experience and to drive consumers to take action. It relies on machine learning algorithms, which analyze all the visitor data on a website, and make predictions about the behavior of each visitor (who could be identified or anonymous, an existing subscriber or just passing through). It then uses this conversion score to trigger personalized actions on the website: this could be changes to content and banners, pushing a targeted offer, or the personalization of messages depending on particular interests or affinities.

2 Content that adapts to readers

In the golden age of the printed press, no one was surprised to see a reader buy a newspaper just to read one part of it. Publishers had no information about the interests of each reader, or what types of stories were most popular. In any case, it was impossible for them to personalize the format and content of the newspaper to each reader. Only a standard version, common to all, was available.

This also meant that editors had to spend considerable time narrowing down the content that went into a physical newspaper, with a set number of pages leading to many potentially interesting stories being left out.

The situation is very different today. The online world gives publishers space to run a much wider range of stories, and by analyzing visitor data through AI they can use this precious information to publish and target content that is always closer to the readers’ expectations.

This capability for better knowing and understanding their readers thanks to AI represents a real opportunity for the media to thrive online.

If this still all seems rather abstract, here are five concrete applications of AI-driven personalization in the media sector:

1. Adapt the experience to the visitor’s interests and preferences

It is now possible to measure every visitor’s preferences and interests in real-time thanks to the predictive algorithms within AI-driven personalization. We can rank these interests for each visitor so as to first and foremost offer them content on the subjects that most interest them, in the format they prefer.

For example, why show football articles first when the reader is there for film reviews? Why favor video content when the visitor prefers written articles?

This could mean reorganizing the content on the homepage of a news site to fit the visitor’s interests. This produces an entirely personalized home page where the content (images, videos, website section headers) appears in the order that best corresponds to the reader’s expectations. And with AI you can do this at scale. By segmenting your audience you can create countless versions of the same page, personal to every visitor.

2. Offer a paid-for article to interested visitors with the aim of converting them

Several studies have shown that one of the key motivations for subscribing to an online news site is its coverage of a very specific subject. This niche content can be extremely well-read and therefore valuable to readers.

Thanks to AI-driven personalization, you can spot when a visitor moves from being a browser to a potential buyer, crossing specific thresholds. So for example, in practical terms, when a reader joins the 20% of visitors most interested in economics, you can automatically offer them a recent, premium article on the subject, free of charge.

If you then deliver a targeted subscription offer to this reader, highlighting deeper content on the subject of economics, research shows that you are substantially more likely to convert them into a subscriber.

3. Increase subscriber numbers by targeting readers at the right time

AI-driven personalization can also target readers at the time when they are most likely to subscribe.

Let’s take the example of Kevin, a young reader who doesn’t actively seek information. He’s happy enough exploring his Facebook news feed and clicking on the articles that interest him. Kevin is a “news bumper”  who consumes information via intermediary sources. He is not particularly inclined to take out a paid-for subscription on any particular news site, even though he regularly visits many.

That said, Kevin often clicks on articles on the same subject: football. He is interested in his team, in their results, in the transfer market and in any rumors about them. This makes him a potential target for a subscription to a website offering specific football-related content, even if he won’t actively go out and start the subscription process himself.

AI-driven personalization can assess Kevin’s level of interest in real-time and offer a subscription at the moment when this has the best chance of converting him. This means he is not bombarded with offers when he is unlikely to buy, and thus also improves the user experience and brand reputation.

4. Optimize advertising by predicting how many ads will be tolerated

Online advertising is the primary revenue source for most media companies on the web. As news sites are often remunerated by the number of impressions or clicks on a given ad, it is very much in their interest to maximize the amount of time ads are displayed and to encourage readers to interact with their content. However, too many ads disturbs the reading experience, annoys consumers and ultimately leads them to leave the website and get their news elsewhere. Achieving a balance between volume and consumer preferences is therefore vital.

AI-driven personalization can precisely determine the number of adverts or the maximum duration of advertising that a  visitor is prepared to see before leaving the website. In other words, it determines their tolerance of advertising pressure.

By assessing the probability of a visitor leaving the website because of too many adverts, you can avoid losing readers, while  precisely delivering  the right number of advertising displays. This maximizes revenues and the customer experience at the same time.

5. Choose the right threshold for the paywalls

Putting in place a paywall is now a common practice to protect content revenues and to encourage people to subscribe to online media. Beyond a certain reading limit (articles or words read in a specific timeframe), the visitor is blocked from reading more content, forcing them to  subscribe if they wish to view more articles.

The real question is: how do you choose the right moment to trigger the paywall? There is clearly a balance between providing sufficient content to get the reader interested, but not too much that means they don’t subscribe. However, typically publishers have adopted a ‘one-size-fits-all’ model, with paywalls triggered at a set point, irrespective of the reader’s personal interests.

AI-driven personalization lets you measure the probability that a visitor will take out a subscription, based on the number and type of articles they read free-of-charge before they have to subscribe. The trigger point can then be set in real-time for each and every reader. This means that the paywall can be triggered at the right moment, maximising the conversion of readers into subscribers.

The move online has radically disrupted the business models of publishers - however by embracing AI personalization they can get closer to their readers and thus build loyalty and greater revenues moving forward.

Topics covered by this article
Abdel El Ouazzani Kameleoon
Abdel El Ouazzani
Abdel El Ouazzani is an AI consultant at Kameleoon, and works with our clients on projects involving AI, machine learning and predictive algorithms. On the blog he shares his expertise and explains the value of AI personalization for specific industries.