Over the past year the Times and Sunday Times newspapers have developed a metric to measure the propensity of its online readers to subscribe to paid digital packages, with News UK tweaking the way it markets to these customers accordingly and boosting conversions by 190% in the process. The typical experience we are talking about here is finding
Over the past year the Times and Sunday Times newspapers have developed a metric to measure the propensity of its online readers to subscribe to paid digital packages, with News UK tweaking the way it markets to these customers accordingly and boosting conversions by 190% in the process.
The typical experience we are talking about here is finding a Times or Sunday Times article online, only to be met by a registration request or, if you had exhausted your free articles for the week, a call to action to subscribe to the paper’s digital package for £6 a week.
Speaking at News UK’s first Change product and design event in Shoreditch this week, Claire Garlick, deputy head of UX at the Times’ publisher, News UK, delved into the Times‘ attempts to personalise this subscriptions process by using a data driven-approach, or what she called its ‘Moneyball moment’.
How they did it
The first step was collecting end-to-end customer data via its own analytics and a set of reader surveys, with the intention of finding out who was visiting the site, where they were from and where they were entering and dropping off.
This data showed that a staggering 96% of users entering the subscriptions process were eventually dropping off. The conclusion of Garlick’s team was that this was because they were catering to one user type, which accounted for a tiny proportion of their readers. So they needed to get more personalised.
The next step was to covert this data into a metric for ‘propensity to subscribe’. “We started to use this data to score based on previous behaviours of similar users and to measure warmness or coldness to subscribe,” she explained.
This led to a redesign of the registration and subscription calls to action on the sight. If a user scored low they would be nudged to register, if they scored high they would be put straight into the subscription purchasing funnel. Either way that user would have less steps to take to get to the end goal for the publisher.
This was rolled out on a small scale to start, for just its Canadian readers and less than five percent of overall traffic.
The first thing they learned was that arbitrarily setting the propensity threshold at 0.5 (on a scale of 0 to 1) wasn’t optimal. So they started to tweak this according to early results to get more sophisticated with their targeting.
“We also wanted to start thinking about different countries and potentially look at different thresholds, to treat them all differently and segment them to find that magic number across them all,” she said.
The project has been running for around a year now and the early results show a 190% uptick in subscription conversions in the six months since rolling out the metric.
“Now we want to build on top of this to push further and personalise more and tease users in with content to warm them up more,” Garlick said.
Data driven editorial
Picking up on the theme, Dan Gilbert, director of data technology at News UK talked about how the media company is working to retain subscribers once they are in the door, with the internal focus being on tools to help editors understand how stories are performing and make more data-driven editorial decisions.
The answer, so far, has been another homegrown metric: ‘Dwell Time Index’.
Going beyond typical customer churn metrics, Gilbert and his team looked to develop a dynamic metric which took into account things like dwell time and unique users visiting its sites, but also adding context for things like article positioning on the page and length.
The reason for this is that dwell time can sometimes become a proxy for article length and traffic can often be driven simply by how an article is positioned by an editor: at the top of the page, for example.
This new metric runs on a range of 0-200, with 100 being an average performance and anything above 180 as performing way above similar articles of its length and positioning.
Gilbert and his team started by giving editors access to this metric via a dashboard, but found that “editors don’t like dashboards”, so they created a rudimentary web overlay to present the stats directly onto the Times and Sunday Times websites. This “started to get the newsroom engaged” and was followed by a more refined Chrome extension, as well as a Slack integration for editors to query the stats directly within the app.
Both of these projects show a concerted effort by the media company to start to take a more modern, data-driven digital marketing approach to grow its subscriber base, the lifeblood of an industry where Facebook and Google has sucked all the value out of display advertising against free-to-access content.