Key to Success:
- The creation of a specific working group to focus on the project was critical to make decisions quickly. We also worked in two-week sprints to rapidly test and learn. With its own internal-facing sub-brand – Project James – the team was made up of experts in editorial, Marketing, AI technology, and data, all working together to take advantage of the new technology.
- The technology itself gave rise to opportunities in content recommendation that until now had not been available. AI technology was at the center of the project, however this was only able to be leveraged through a series of key data initiatives.
- The project itself provided real benefit to consumers who want to receive as much value as possible from their subscriptions. A news summary email based on previous history strengthens the Time proposition by making news content more relevant, in the right format, at the right time, and with the right frequency.
The Times provides news content from behind a paywall, therefore driving value from the subscription is key. Content consumption forms a key part of the engagement strategy. The challenge we faced was to use AI-driven models to build engagement from the newsletters, thus building out customer value and reducing churn.
A rigorous and innovative testing strategy was key to bringing the strategy to life.
First, six different core content propositions were explored: Daily Briefing, Cartoon of the Day, Top Commented & Top Shared, End the Day on a High Note, Suggested Reading, and top 20 articles. Together, these make the core components of the Times news summary emails, however different readers all consume the content in their own order. Eleven models were built that contained various combinations of the propositions, and then these were tested for optimum impact.
Next, time-of-day and frequency variables were tested to understand how these factors can be varied to optimized open and click rates.
Finally, formats were tested; the design and layout of the emails were created using different combinations of copy length and number of images, to see which formats different consumers preferred at different times.
Using AI technology to build and then evaluate the efficacy of each of the variables resulted in the optimum combination of content, timing, frequency and format, with the most successful variables being used more and more.