A dive into personalisation isn’t a small project and will take a great deal of effort. Now I imagine you are wondering where to start. Depending on your appetite for risk, you will need to find the right place to start running some tests.  

There are two schools of thought as to where to start: 

1. Next read/article pages

I am going to call this the cautious approach. At the end of most articles/content pages, you will most likely have a section that has recommended reads. For most I suspect this is automated; for some it may be handpicked. The recommendations are likely based on the topic, author, and content itself rather than who is reading it. Users expect automation here, and editors rarely expect to have huge input. That’s why it’s a good place to quietly start without needing to get many different departments involved. 

A great example of this is NZZ, which got an honourable mention in the INMA Global Media Awards. As you’ll see from the image below, they redesigned their next read section. The top three articles remained in the previous format and below that two sections are personalised. In this process they also changed design. The results are staggering: 40% average CTR uplift and 63% average completion uplift. (If you want to learn more about this, check out the recording of the Product Initiative Meet-Up here.)

2. Homepage personalisation

I am going to call this one the headstrong approach. Overall it’s likely to be the most-viewed page and there will be a lot of stakeholders with a lot of opinions. You get the biggest bang for your buck, but you don’t want to make any major screw-ups. To be fair, there are some ways to limit this, and most news organisations test with a small section of the homepage, likely below the fold.

Mediahuis Netherlands ran some tests to personalise specific segments of the homepage in a way that was palatable for all involved. This approach minimises risk and allows the team to gather some results.

And if you want to go all in, I highly recommend looking at what Schibsted’s Aftenposten in Norway has been doing. They are more advanced than most with their homepage personalisation. Below is an overview of the structure they use. Notice the mix of editor selection working alongside the algorithm and how they use sub feeds and bundles to group content to ensure there is a mix of content on the homepage that truly represents Aftenposten’s brand. 

Their results so far are impressive: 

• By automating the feed and removing read articles, they saw an 8%-10% increase in CTR. 
• They saw a 30% increase in CTR by retargeting users interested in games. 
• And when they have used their “conversion maximising ranker,” they have seen a 20% lift in subscription sales. 

For most companies I would suggest starting small, experimenting and building on what you learn.

If you’d like to subscribe to my bi-weekly newsletter, INMA members can do so here.

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .