This blog post was co-written with Trent Woodbury, a Data Scientist at Pagos. Specifically, it focuses on the unique challenges news organisations face in engaging readers through push notifications and emails.
This is the first in a series exploring these challenges. Future posts will focus on specific use cases with actionable strategies to improve personalisation and engagement.
Setting the Stage
In the attention economy, companies compete relentlessly to capture users’ focus. Attention is a finite resource, and people can only process a limited amount of information (Nielsen Norman Group, 2019). However, capturing that focus has real monetary value, ranging from advertising revenue to projected profitability based on Daily Active users (DAUs) count.
Social media platforms like Instagram and disparate app push notifications are prime examples of tools designed to drive user engagement. But as more companies push for attention, there’s a real possibility that users will suffer from notification fatigue. The result? Disengagement and churn (MUO, 2021).
For product teams, there are different ways to tackle this problem. One is to turn over-communication into a constraint. Limitations like “one email every X days” can be imposed to avoid spamming users. But there’s another way to think about this:
- What’s the cost of spam?
- What’s the worst case scenario?
Well, the risk is that your users may unsubscribe, leading to churn and impacting Life-Time Value (LTV), which is your company’s estimated total net profit that you can expect from a customer throughout their entire relationship with you.
However, the key isn’t sending fewer messages. It’s sending the right messages at the right time. In this post, we’ll walk you through how news organisations can personalise notifications and emails to boost engagement and net revenue.
Why Aren’t Your Readers Engaging with Push Notifications and Emails?
Unlike entertainment apps that deliver escapism or social media platforms designed for endless scrolling, news outlets depend on delivering timely and relevant information. However, look at any user’s phone and we can guarantee you that their screens are inundated with alerts about new messages from friends and families, TikTok trends, or Instagram stories. Resultantly, poorly executed notifications will often go unnoticed by the user or even drive users to turn them off altogether.
For news organisations, the problem often comes down to two key issues:
- The push notification is irrelevant to the user. For example, if the user has never shown interest in Italian politics, sending a notification about Giorgia Meloni’s new immigration policy will have a negative estimated ROI for the customer.
- The push notification is sent at the wrong time. First, notifications need to be sent on localised times, not global times. That way, your readers in South Korea won’t be getting notifications at 3AM. The best way to solve for this is to alter your data to be localised: always send based on the user’s time zone. You can also analyse what times have the highest/lowest open rates and only send during the hours with the highest open rates. Just make sure that when you’re analysing hourly open rates, you’re looking at localised hourly open rates. If you have significant presence in multiple different countries, you can also chart open rate by hour per country, since countries like Spain tend to have different schedules than Japan.
Both of these challenges highlight a deeper issue: in a crowded attention economy, news organisations can no longer afford to send generic or mistimed notifications. To compete with entertainment and social media platforms, every interaction must feel relevant, timely, and valuable.
What Does Good Personalisation Look Like in Practice?
In a previous post, we explored Product-Led Growth (PLG) strategies for acquisition and retention. A product’s value proposition is central to its market positioning, and personalisation plays a big role in delivering that value. It further boosts LTV.
Data, naturally, is the foundation of effective personalisation. It can come from disparate sources, like reading history, click-through rates on articles, or even The New York Times’ For You feed. These data sources work together to identify user preferences. It can signal whether a reader is more interest in sports updates, political analysis, or articles about new technology. As an example, if a user frequently reads technology related articles, recommending a related in-depth report or sending an alert about breaking news in this area can drive higher engagement.
Recurring usage patterns are another strong signal of value. A reader who opens your app every morning to check the top headlines is showing loyalty. Another example is personalised emails.
Personalised emails to increase open rates and click-through rates
Personalised emails should have two primary features:
- Relevance: the email’s content should reflect what the user has engaged with recently. “Recently” here can mean over the last 30 days, 90 days, or even year. The exact parameters for recency are something that you should establish through AB testing: you can test the same personalisation algorithm but with user data from the past X days as input. You can also bias more toward more recent content by weighting recent content interactions more strongly than content that the user engaged with 70 days ago.
- Variance: since an email can link to multiple different articles, we want the articles present in the email to reflect the largest possible range of customer interests. Think of this as a shotgun approach: the customer has shown interest in international politics, AI development, and European fiscal policy. A personalised email should have links to articles in all of these topics. Think of this as a strategic shotgun approach.
How Can Personalisation Improve Reader Retention?
- Predictive personalisation model to match readers with relevant content. Personalisation allows you to show the reader that your news site contains articles that tailor to their interests.
- Balance curated content with editorial judgement and data-driven recommendations (risk: filter bubbles). Editorial judgement shows that you have journalistic integrity and are investing in reporting that your newspaper believes in. The Financial Times balances this nicely with a “For You” personalised feed alongside a universal feed determined by their editors.
What Role Does the Product Manager Play in Personalisation?
The Product Manager is responsible for:
- Defining the problem. If you don’t have clear outcomes you want to achieve with personalisation, it’s unlikely your efforts will deliver value. Start by understanding the problem you’re solving and the impact you hope to create.
- Connecting it to business goals. Make personalisation part of a clear vision and ensure it align with your business objectives. Assess its feasibility (technical constraints) and viability (ROI potential) before investing resources. Given limited engineering capacity, trade-offs are inevitable, so the Product Manager must balance priorities to decide where personalisation fits.
- Focusing on user needs. Personalisation must address what users need and are willing to pay for. In the journalism context, your product serves more than just functional tasks (staying informed). It also supports users’ social and emotional aspirations. For example, users might want to share relevant content or join discussions (social) and feel empowered or reduce anxiety about misinformation (emotional). Map these jobs to understand personalisation’s role in improving the user experience.
- Lead cross-functional collaboration. Successful personalisation requires buy-in from multiple teams, including engineering, data science, and stakeholders like editorial teams. It’s the Product Manager’s job to align everyone around the strategy and drive implementation.
If your organisation is new to personalisation, cross-selling the promised land of its benefits is key. Start small, using experimentation to test and validate ideas, and build internal confidence. For example, hypothesise that personalisation will boost retention based on specific challenges you have identified. Create an AB test applying the personalisation to a small percent of your customers to measure the impact, then scale up to all customers if successful. Ensure that the test is big enough to establish statistical significance within a reasonable time frame. Your Data Scientist can help you perform a power analysis to determine how long the AB test will need to run and how many customers will be needed to get statistical significance in a given time horizon.
At any rate, avoid introducing features without first anchoring them in user needs. Applying practices like Continuous Discovery here ensures rapid iteration and helps unlock meaningful results.
What Role Does the Data Scientist Play in Personalisation?
The Data Scientist is responsible for:
- The analytics around user engagement that can drive personalisation.
- Development of the personalisation algorithms. This includes the metrics used to evaluate the performance of the algorithm. Determining these metrics should be done in collaboration with the product manager.
- Managing and evaluating AB testing of different personalisation algorithms. Data Scientists should be aware of which changes to the algorithm lead to improved engagement.
- The data scientist is also responsible for working with Product to aid in data driven decision making: e.g. when improvements to the email personalisation algorithm have shown diminishing returns in AB tests and it’s time to shift focus to improving the push notification personalisation algorithm.
What Could Go Wrong?
- Optimising for click through rate often leads to an emphasis on highly negative content. This is because negative content tends to have higher engagement in general. This comes in direct clash with journalistic integrity, though. If your responsibility is to produce an informed citizenry, only notifying users of the worst events in their community/world produces a caustically misinformed customer base.
- Similarly, ignoring emerging interests misses broader engagement opportunities. Personalisation based solely on past data points can overlook trending or high-impact topics, like global events or breaking news outside a user’s usual preferences. This also limits engagement and risks leaving users uninformed. This can be overcome with an algorithm that allows for a certain amount of randomness to be introduced into the recommendations.
- Algorithmic bias can reinforce inequality and alienate audiences if managed poorly. They may favour certain demographics, exclude underrepresented groups, and damage trust in the organisation. As an example, an algorithm prioritising clicks might push NFL stories whilst ignoring women’s sports or local events, leaving these topics and audiences overlooked.
- Overpersonalisation, especially without users’ consent, is real and can be perceived as creepy. Give users the option to consent to personalised feeds. Likewise, users should be able to easily opt out at any time.
What Should News Organisations Do Next?
News organisations need to blend personalisation with broader discovery efforts. Use data to surface both what users care about and what’s trending, so they stay informed about familiar and important global events. This avoids overly narrow recommendations and keeps engagement high.
Equally important, editorial teams should assist product teams to tailor personalisation efforts that ensure quality and trust. By combining personalisation with strong editorial oversight, news organisations can create engaging and reliable experiences for their readers.
Final Words
Personalisation works best when it’s balanced with discovery, fairness, and editorial oversight. By focusing on trust and value, news organisations can engage readers while keeping them informed.
Leave a comment