Audience Engagement Digital Publishing
2 mins read

How publishers are using personalization

Getting your Trinity Audio player ready...

There is more than one way for publishers to personalize their digital products for users

Personalization has been a buzzword in publishing for several years, but there isn’t one, correct way for publishers to personalize their digital products for users. Writing on the Product Initiative blog for INMA (The International News Media Association), Jodie Hopperton has identified  key trends to watch in the way publishers are using personalization.

Takeaways

  • Jodie Hopperton is a US-based media executive who has worked for several international news organizations. She is currently Product Initiative lead for INMA, where she has turned her attention to personalization in publishing on the Product Initiative blog.
  • Hopperton acknowledges that publishing’s past was firmly focused on serving up content that publishers, editors and journalists thought audiences should be reading. While that’s still part of the job, the dangers of information overload mean modern publishers should help to narrow users’ choices.
  • She identifies two approaches that publishers can take to help filter down audience choice. These are active personalization, relying on publishers asking users questions about themselves; and passive, using technology to make recommendations based on demographics, past behaviors and other data signals.

Active personalization

  • The questions asked in active personalization range from simple location information to more complex questions around subject preferences. Gathering user data can be part of the registration process or driven by activity, frequency or time-based prompts.
  • Based on her own research, Hopperton says users who actively select their preferred topics of interest display significantly higher engagement. She points to Spotify, Substack and Apple News as service providers that use active personalization well and how they deploy it.

Passive personalization

  • Passive personalization infers user preferences by combining established publishing signals – content value, timeliness, author – with data signals based on behaviors. It can be as complex as the publishers infrastructure will support, but Hopperton gives the example of the four signals used by The South China Morning Post:
    1. Location
    2. Behaviour
    3. Loyalty
    4. Persona
  • Hopperton says that TikTok is arguably the best company in the world at passive personalization. She notes that, at no point in registering for the app or in using it does the user make any choices about what they want to see. Instead the app infers viewing preferences based on data signals captured from previous behaviors.

Personalization trends

Personalization trends that publishers should pay attention to include: 

  • Indicating to registered users or subscribers that you know who they are by using their name.
  • Rewarding subscribers with page flashes or badges to differentiate subscribers from regular visitors.
  • If you are using active personalization, add a “For You” or “My News” tab to show users where their content is.
  • Helping people breakout of their own echo chambers by ‘counter-personalizing’, suggesting content that they might not normally read.

Hopperton thinks that as more publishers personalize, the use of personalization techniques will evolve. For example, standardization could help users navigate across brands and services. But she says, as publishers learn more about their own users, they should innovate within their own properties. She says: 

We can experiment ourselves and look to create moments of delight that they are not expecting.

This piece was originally published in Spiny Trends and is re-published with permission. Spiny Trends is a division of Spiny.ai, a content analytics and revenue generation platform for digital publishers. For weekly updates and analysis on the industry news you need as a media and publishing business, subscribe to Spiny’s Trends weekly email roundup here.