Newsletter: A/B testing and video study

Hi there,

At smartocto, we are constantly driven by your questions, and one query that keeps cropping up is about the impact of editorial actions. Take, for instance, the simple question: if I embed a video in an article, does it boost engagement?

This is a fantastic query at the intersection of data and journalism because it comes with actionable insights. Over the European summer months, we've been hard at work and have curated a series of articles that will provide answers to such questions: From Our Data.

One particularly intriguing question that we haven't added to this series is about the positive effects of A/B testing headlines. Curious about the findings? Well, we've just announced a free webinar where we'll spill the beans.

What makes this webinar exceptional is that, in recent months, we collaborated with a client to explore the use of ChatGPT in generating alternative headlines. Is AI a game-changer? The answer awaits you in this exclusive webinar, and you can secure your spot right now.

Adding videos to articles is not your silver bullet

The first blog from our 'From Our Data' series is already live. We've analysed over 4,000 articles across five smartocto clients, spanning a two-month period. The outcome is significant because, until a few years ago, the future of journalism seemed heavily invested in a 'pivot to video' strategy.

However, now we know that video isn't the silver bullet. This isn't to say that video is irrelevant in online journalism, but our conclusions point to some key insights:

• There's a minor difference in engagement scores between articles with videos and those without.
• This difference can be attributed partly to distribution strategies, such as push or no push.
  
• The type of video format used within articles matters; stories told primarily through video tend to garner significantly higher audience engagement.

Feel free to dive into the data and peruse the tips provided at the end of the blog. We invite you to draw your conclusions based on this case study.

In our analysis of video impact, we rely on smartocto's Content Performance Indicator (CPI). The CPI score is a compound metric that factors in various measurements, including page views, attention time, read depth, and social actions, among others.

While there's an overall CPI score, we also offer three additional CPIs that provide insights into specific reader behaviours:

• CPI Exposure: Reflects the content's reach, including metrics like page views and social actions.
• CPI Engagement: Offers insights into visitor actions while consuming content, such as attentive reading, time spent, thorough reading, and more.
• CPI Loyalty: Provides information about the behavior of highly engaged, habitual readers, who are typically the most valuable audience for media.

To make this complex topic more digestible, we've compiled five questions and answers in a blog post. This includes why we consider CPI a robust metric and why you can trust the algorithm behind it.

Webinar

We don't want to jump to conclusions prematurely about our ongoing experiment with Omroep Brabant. However, given the scrutiny on the relationship between journalism and artificial intelligence, we're confident that our upcoming webinar will be of great interest to media professionals.

What language models are primarily good at is rephrasing existing texts. With a simple prompt, you can have your own pieces rewritten by J.K. Rowling or Nelson Mandela, but in our view, it's more useful to ask ChatGPT for alternative headlines. Our client Omroep Brabant has taken up the challenge, and we will discuss the outcome in the webinar.

Save the date for our webinar and be at the forefront of the latest developments in journalism and AI. We can't wait to see you there, so sign up!

Reading tips

• Micropayments for news have not proven to be a successful model for Dutch media platform Blendle, as it pulls the service for German and US audiences in September. Niemanlab points out there's been very limited demand and the rise of digital subscriptions have done the idea in.
• News outlets are experimenting with AI-generated content. Meanwhile, a coalition of major publishers is considering a lawsuit against AI companies for content theft. This highlights the media industry's uncertain response to AI disruption, NY Mag writes.
• Over and over again media found themselves reporting about things happening to Twitter (now formally X) which weren't executed at all. The press should take a more skeptical approach. It's time to change how we cover Elon Musk, Casey Newton makes his argument on Platformer.news.
• Young people function on different frequencies than other generations when it comes to news, so the best way to counter disinformation amongst them is to keep them better informed, in their style, Diana Filimon says at The Fix.

Thanks again for reading this newsletter. We'd like to remind you that there is a product newsletter and a webinar newsletter as well, to make sure you won't miss any new webinar. You can subscribe via this link. Or ask for a free demo if you'd need to see what we've got to help your organisation forward. 

See you next time!
Team smartocto