Predicting Social Media Engagement for News Articles & Digital Content


It is becoming increasingly important for media companies to deliver content that strengthens engagement with their audience.


Online global monthly unique visitors


Subscribers to three of the leading digital newspapers in the United Kingdom


Increase in social engagement and increase in paid subscriptions for digital news


Our client is one of the largest media organisations in the UK. They wanted to find a data driven solution to help their newsrooms make better informed decisions about which articles to distribute across social media channels in order to increase social engagement.


Our team of data scientists were given access to minute-by-minute social media traffic data for all news articles and content the company had created in recent months.
The first goal for the team was to create an accurate prediction model for the total number of visits any given article could achieve within the first 24 hours of publication. A second goal was to predict social media traffic on any given day at any given hour. The ultimate aim was to find out what news articles and content to create and when to release them for publication.

press to zoom
press to zoom
press to zoom

In order to predict the total number of views an article would achieve in the first 24 hours the team used a combination of statistical text mining, Natural Language Processing (NLP) and regression techniques to understand which topics received the largest number of visitors and generated the highest levels of engagement with detailed insights into the title of the article, word count and the timing of the publication. Prediction of social media traffic was achieved by using classification and regression tree algorithms to understand what causes the highest levels of social media traffic on any given day at any given hour across different social media channels.


By using a combination of NLP, classification and regression models the team created a solution that helped our clients analyse an article and predict the total number of visitors it would receive. The results delivered a 28% increase in social engagement and subscribers to digital content. Since then the business has delivered exceptional growth in revenues, profitability and customer engagement.

Note: Delivered in partnership with Pivigo Limited.