Why personalization isn't a silver bullet for publishing & media

< Blog home

Tailoring to users helps, but giving them agency and being unique are more crucial to your success

With so much content so easily available to everyone online, it’s hard to hold people’s attention. Helping your users to find your most relevant content is essential if you’re going to do that. Nowadays, many efforts to do this are lumped under the umbrella term 'personalization’.

To call it a hot topic is a bit of an understatement. In a recent survey done by DigiDay, fifty-six percent of publisher executives said personalizing of content is “a high or very high priority”.

Alan Soon
Alan Soon, co-founder, The Splice Newsroom

Many commentators, such as Alan Soon at The Splice Newsroom, describe personalization as the future of media. Some have even started to score companies based on their personalization efforts. Last year, Sailthru, an experience management platform for retail and media, launched its first annual Personalization Index. Currently they’re only focused on retail (Sephora ranked #1 on the top 100 ranked list, in case you were wondering), but the lessons learnt almost certainly apply generally.

There were three takeaways for businesses:

  1. A personalized user experience is a higher quality experience
  2. Focus your personalization strategy on what makes you - and your users - unique
  3. Don’t focus on just one channel, but strengthen the user experience across all touch points

Silver bullet?

Does this mean the verdict for improving your user experience is definitively in? Can we all conclude that more personalization is the way forward, and that the forty-four percent of publishers who say it’s not a top priority are missing a trick?

Not necessarily. It's important to recognize that the situation for Sephora is different to that of a publishing or media business. There are good reasons why retail brands star in this new Personalization Index: publishers and media have different products and models. For a start, online retailers often have stronger data signals from their users in the form of purchase histories.

Personalization is often presented as a silver bullet to solve all your problems in audience engagement and retention, but let’s take a look at the ballistic effectiveness in the publishing and media universe.

What makes you, you?

Firstly, we can agree that some of the content you create, all of your users should see. I’m talking about that groundbreaking investigative story, exclusive partner content or flagship new course.

It’s The New York Times’ Harvey Weinstein Paid Off Sexual Harassment Accusers for Decades, Netflix's Stranger Things and your own defining content that embodies what your site is all about. Poorly implemented personalization will hide this from your audience.

Mike Dyer
Mike Dyer, former president and publisher, The Daily Beast

You need to be aware of your mission as well. “Journalism companies have a public good mission to confront people with fact-based reporting regardless of how they feel about it”, said Mike Dyer, former president and publisher of The Daily Beast, in a 2017 interview. “[T]oo much personalization or content matching can harm that mission.”

For the news media, their editorial choices are part of their product. Moving away from what makes you stand out is likely to turn off users and commoditize your offering, as The New York Times found out and I discuss below. 

This is absolutely connected to Sailthru’s second takeaway about focusing on what makes you unique. It’s important to have your brand shine through remarkable, distinctive content. You should also consider whether you have the responsibility to deepen and broaden the perspective and experiences of your users. Personalizing the entire content experience can build echo chambers.

Going halfway

A second thing to consider when you're planning on more personalization is whether you actually know enough about your users to make a more personalized experience possible? As Andrew Stothert, CEO at Brand Vista writes on the topic: “Without the full picture, you are in danger of making stuff up. A strategy that will never bode well with something as business critical as the customer experience.”

Andrew Stothert
Andrew Stothert, CEO, Brand Vista

Don’t force people down a siloed content path based on limited data from a single source such as browser history, location or device type. If you don’t have enough information about your user, don’t make the mistake of narrowing their whole experience from the get-go.

Besides, sometimes it's simply not possible to tailor your content to each individual user’s attributes. This can be down to your monetization model, the tech situation or philosophical and ethical considerations. In the survey I mentioned, Digiday discovered that more than half of publishers can only adapt their content to the page context, and not to individual users.

A user also doesn’t wear the same hat every time she or he visits you. It’s important to realize than an individual has different personas, with divergent goals and preferences. Jane at 10pm on a weekend doesn't want to see the same content as Jane at 10am on a Monday. Clumsy personalization conflates these personas and can actually harm your attempts to show people the right thing.

If you’re in a situation where individual users’ attributes are not available, don't despair. There are still many ways to make sure that you serve relevant content and optimize the user experience.

Making the most of every page

One way to optimize without needing to personalize is to display on-page content recommendations. These can be based on the content and context of the page. With around 80% of external traffic to sites coming from search and social, it's important that your article pages can show off all the other great content on your site.

Picture a side panel that allows for more ‘diverse’ discovery throughout your site - e.g. a way to explore a wider range of topics. It’s for users who didn’t make it to the end of the article but could be interested in one of your trending or evergreen articles. You could also offer end-of-article recommendations that enable deeper discovery into that particular topic. And what about inline recommendations that for example focus on a specific organization, geographic location or topic mentioned in the paragraph above?

Also read Making the most of every page

Another way to optimize without personalizing is to make recommendations based on the behavior of your audience as a whole, rather than individual users. Maybe Jane at 10pm on the weekend is more like Jim and Angela at that time than she is like herself on Monday at 10am. Look at what the rest of your audience likes at that point and you might just find the best thing for her.

This also works because studying audience level data gives you an insight into the journeys your users are taking through your content. The aggregated data can show you the pathways which users enjoy the most, which can then be replicated in the suggestions the recommender system makes. This audience-based approach can be very effective even where you lack the data required for effective personalization.

Optimization at scale?

Selecting, delivering and continuously updating the right content recommendations is a difficult and time-consuming task. If you consider the amount of content that publishing & media sites produce, and changing audience behavior on top of that, then it’s clear that for a human this is going to be a hugely difficult task. A good content recommendation system that can analyze all content and update suggestions as new content is added is an obvious solution.

Data Network Effects

As well as using semantic technology to understand the meaning and context of a piece of content, the best systems are driven by machine learning. This is what enables them to learn from anonymous user behavior and take the audience-based approach that I've set out above. It's a scalable, algorithmic approach that can throw up surprising results that a human would never intuitively pick.

Your personal way forward

The third and last point I want to make is on the potential disconnect between personalization and your users’ agency. Do your users actually want to see more and more targeted content?

When The New York Times announced their plan to further personalize their site and apps in March last year, there was an uproar among readers. More than 300 comments came in (before they closed down the comment section), and the vast majority were against indulging readers rather than giving them the benefit of the Times' editorial expertise. The following note was voted ‘top’ comment:

Top comment from the backlash against The New York Times' personalization plans

I recommend you go through the Readers’ Picks and read the passionate replies, such as this one on the importance of individual agency and bursting your filter bubble:

A Readers' Pick comment from The New York Times' personalization plans announcement

Handle with care

Personalization can be a powerful tool, but it's unlikely to be a silver bullet for user engagement challenges at the moment. Here are some thoughts to summarize:

  • Investigate whether your audience has an appetite for a more tailored experience and if so, where on your site?
  • Think carefully about how much of the experience you want to personalize, and based on which data?
  • Strengthen the whole experience and always offer different discovery routes on your pages - don't assume you know what your audience wants. Try a variety of approaches, such as offering suggestions based on similar article context or evergreen status.
  • If you move forward with (more) personalization, then a recommender system that adapts to changing goals and preferences is a must. If you don’t move forward with personalization, an adaptive system should still be part of your user experience plans.
  • Be transparent with your users about why they're seeing what they're seeing, and give them the power to influence it if you can - perhaps they actually want to see what you think is important, and not what you think they might want.

Recommended for you

Interested in this topic? Get in touch to talk more