How the FT is working with :CRUX to develop a knowledge score for readers
By James Webb
Building knowledge is a huge motivator for FT readers. Our customer research suggests that finding context; knowing what they should read next, and why, can often be a challenge. How do they know their reading has an impact on their knowledge?
With DNI funding from Google, we are excited to be partnering with :CRUX, a technology company dedicated to Quantifying Knowledge, to provide our readers with a live knowledge score and suggestions for what they should read next.
It works by measuring how much users might know based on their reading behaviour, and recommends the best content to increase their knowledge. We think this provides an opportunity to deliver a powerful behavioural hook, with strong potential to grow reading habits among both our lower engaged subscribers and super fans.
At the FT, we have an engagement formula which has become our ‘North Star metric’ in recent years. We measure many of our engagement initiatives against this metric, as it is a direct indicator of a subscribers likelihood to stay with us. Subscribers who are more engaged have a higher LifeTime Value (LTV), so anything we can do help them discover more relevant content; to visit more often; and to consume more stories per visit, will have an impact on their engagement score, and on subsequent revenue.
With :Crux our aim is to measure whether using this system has as powerful an impact on user engagement as our initial research suggests.
How it will work
:CRUX measures the knowledge contribution of an article to a user in the context of a specific topic. Users will see their knowledge score grow as they engage, and will receive a list of recommendations for the best articles to build up their knowledge. Additionally, they will see what happens to their knowledge when they’ve let go of a subject for a while. And, we hope, we’ll be able to give them the option to receive alerts about changes in their knowledge levels due to articles they missed.
What makes this interesting?
At the FT we’ve experimented with numerous ways to build user engagement, both through customisation and personalisation, and more recently, by tailoring individual user journeys. For instance, what might a user coming from a myFT digest email see on an article page, that someone coming from the homepage doesn’t?
Taking a user journey based approach is a key focus for us, but this offers a unique angle on personalisation, which we hope will be compelling for manys of the FT’s subscribers. With this in mind, there are several aspects that make this project interesting:
- It visualises and gamifies and gives the user control on a major motivator for consuming news — Knowledge. By letting users see the positive results of their reading we believe they will read more, and perhaps as importantly will be more satisfied with what they read anyway.
- It creates a whole new incentive to click on an article — not only because you’re interested in the headline, but also because you know how much it will contribute to your knowledge.
- It creates a way of maintaining engagement with topics in a way that’s friendly for the user’s schedule — through our knowledge-drop quantification feature. It helps users not to lose what they’ve already gained in terms of knowledge if they let go of a topic for while.
Shining a light on our metadata
One of the other key aspects that makes this project interesting is that it shines a light on our metadata powered approaches. While this is by no means the only factor in delivering content relevance, it is still a major consideration for how we structure and deliver meaning to articles at the FT.
Just another recommendation engine?
There are numerous companies that offer recommendation systems, often optimised for specific niches. Meanwhile, some publishers have explored their own bespoke solutions, which seek to make best use of reader-based (behavioural) signals or those exclusively powered by metadata. Some systems make observations and recommendations about similarities in content consumption between individuals and user groups (eg collaborative filtering). These systems are effective but by definition they don’t fit methodic knowledge building about a specific topic and act more as ‘serendipity engines’.
None of these examples focus on knowledge as a valuable asset to be acquired or use Knowledge Acquisition as an engagement hook. While we will continue to experiment with other approaches to content recommendation, we think that Knowledge Acquisition is a one of a kind tool being utilized by the FT. It also aligns with our product vision, which is geared around providing readers with the most relevant information, in part by surfacing key topics of interest.
What do we hope to learn?
At the end of the project we hope to have gained insights into which type of subscriber finds this recommendation the most useful, while also learning which touchpoints prove most powerful. More broadly, we hope to learn the following:
- What patterns might emerge around what topics people find most appealing to this tool, and to what extent does this offer a useful feedback loop for our Editorial team?
- Do users react more strongly to knowledge incentives around topics they are more or less informed about?
- Do users respond better to positive knowledge feedback instances or negative ones?
If successful, we hope to be able to introduce the concept not just to subscribers, but to anonymous users as a way to introduce them to FT content in an entirely different way.
We’ll post updates later in the year to keep you informed on how the project progresses.
