Through dynamic pricing strategies, health score-based assessments and personalized retention tactics, Hearst US optimizes subscriber experiences at every touchpoint. By meticulously measuring success metrics and embracing continuous innovation, Hearst US sets the standard for leveraging data-driven insights to foster long-term customer loyalty and mitigate churn in the digital subscription landscape. 

Good morning, Kelli. Thank you for taking the time to speak with us today. To begin, could you provide an overview of your role and responsibilities at Hearst US?

Kelli Dakake: Certainly. I oversee the customer journey from lead generation to engagement and retention, specifically focusing on non-subscribers and subscribers alike. My team works to create habits for customers and increase their engagement across various platforms, while also implementing strategies to prevent churn.

That sounds like a multifaceted role. What would you say are your primary key performance indicators (KPIs)?

My main KPI revolves around retention and engagement. We’re constantly striving to keep our customers engaged and satisfied to minimize churn.

Your work on dynamic pricing is quite intriguing, especially considering its relatively novel application in Europe. Could you elaborate on how you’ve implemented segmented pricing strategies in your organization?

Absolutely. We utilize dynamic pricing across three main points: 

• Segmented pricing at the point of acquisition
• Subscriber health score-based pricing
• Tailored pricing for more tenured subscribers. 

Each strategy aims to optimize pricing for different segments of our customer base, enhancing both acquisition and retention efforts.

It’s fascinating to see how data-driven approaches like dynamic pricing can impact customer behavior. Could you explain more about your subscriber health score and how it influences your pricing strategies?

The subscriber health score is a machine-learning model developed in collaboration with Mather Economics and our in-house data science team. It evaluates various transactional and behavioral variables to assess the loyalty as well as churn levels of subscribers. We decided 75 variables to build it, each of them having a more a less strong impact on the final score.

The main ones are:

Variable	Increase or decrease Churn
Web visits frequency	Decrease churn
E-edition visits	Decrease churn
Newsletter signups	Decrease churn
In intro period	Increase churn
Newsletter sent	Decrease churn
Newsletter clicks	Decrease churn
Web scroll depth	Increase churn
App visits	Decrease churn
Multi device Flag	Decrease churn
Primary device Mobile	Increase churn
…

Based on this score, we adjust pricing to incentivize retention and prevent churn.

> To read next: determining the optimal price for your subscription product

How do you tailor your engagement and retention strategies for different segments of your customer base?

We categorize our subscribers into low-risk, medium-risk, and loyal segments based on their health scores.

Each segment receives tailored communication and pricing strategies. For example, we’ve found that highly loyal customers may prefer minimal communication to maintain their routine, while at-risk customers may benefit from targeted engagement efforts.

Another example: after 6 to 9 months of subscription, we price subscribers based on their current rate as well as their subscriber score.

You mentioned using dynamic pricing across three main points: segmented pricing at the point of acquisition, subscriber health score-based pricing, and tailored pricing for more tenured subscribers. Could you walk us through how each of these strategies works in practice?

Segmented pricing at the point of acquisition involves customizing initial offers based on factors such as geolocation and engagement levels. For example, a potential subscriber who has shown high engagement with our content might receive a more premium offer compared to someone who has shown less interest.

The subscriber health score-based pricing is a bit more nuanced. This score is derived from a machine learning model that analyzes various variables, such as frequency of engagement, platform usage, and response to pricing changes. Based on this score, we determine whether a subscriber is at low, medium, or high risk of churning. If a subscriber is deemed high-risk, we may offer them a discounted rate to encourage retention.

Finally, our tailored pricing for more tenured subscribers involves adjusting subscription rates based on loyalty and engagement levels. 

> To add to your reading list: From newsletters to predicting possible cancellation: tactics to reduce churn from The Audiencers’ Festival Madrid

Could you provide an example of how you’ve used dynamic pricing to prevent churn among your subscribers?

Of course. Let’s say we identify a subscriber with a high risk of churn based on their health score. Instead of waiting for them to cancel their subscription, we proactively offer them a discounted rate or a special promotion to encourage them to stay. By adjusting pricing dynamically based on their likelihood of churning, we can effectively reduce churn rates and retain more subscribers over time.

It sounds like you’re leveraging data in innovative ways to optimize both acquisition and retention efforts. How do you measure the success of these dynamic pricing strategies?

Our success metrics vary depending on the specific strategy, but we primarily look at metrics such as churn rate, subscriber retention, and revenue growth. For example, if we see a significant decrease in churn rates among high-risk subscribers after implementing dynamic pricing, we consider that a success. Similarly, if we observe an increase in subscriber engagement and revenue following targeted pricing adjustments, we know we’re on the right track.

That personalized approach seems crucial for maintaining customer satisfaction. Are there any new projects or initiatives on the horizon for your team?

We’re constantly exploring new strategies to enhance customer engagement and retention. Currently, we’re focusing on refining our segmentation models and further integrating subscriber health score into our strategies. Additionally, we’re exploring opportunities to leverage machine learning for propensity scoring to predict customer behavior more accurately.

1. Gannet: article summarization and daily horoscopes generated by AI

Gannett | USA Today, one of the leading publishers in the US, started shaping their AI use last year, rolling out two different use cases: 

• AI-generated bullet points. These offer a preview to the user as “a way to drive interest, and have the user enter the conversion funnel”, explains Amelie Sutton, product manager at Gannett | USA Today. They are reviewed by editors before publication however

• Daily horoscopes. What started as an article that had to be reviewed and updated daily, is now a template which natural language generation transforms into an article. Suttons shares how “it’s a very low-risk effort, we are only using data from a third-party provider. We use the same template for the lottery, and right now, it’s fully automated. It’s pretty successful and something that people search for”.

2. NewsGames: building quizzes and crosswords with AI

Since the rise of gaming apps (think The New York Times, Apple News and The Telegraph), puzzles, crosswords, and other traditional media fixtures have gained renewed prominence in publishers’ strategies.

But how can local newsrooms, who don’t have the ability to dedicate teams or invest heavily, capitalize on this trend?

Meet Bill Miles and Kevin McCurdy, a pair of tech entrepreneurs, who founded NewsGames, a platform dedicated to crafting puzzles, quizzes, and crosswords tailored specifically for local newsrooms. Their mission? To engage audiences, reduce churn, and create a strong experience around local news.

Utilizing GenAI, the platform enables newsrooms to effortlessly generate quizzes and crosswords. These interactive games can seamlessly integrate into any website or newsletter. 

Each game is meticulously customized for the local newsroom. The only requirement being to input article URLs, images, podcast links, and other content elements. Newsrooms can personalize the appearance by adjusting logos, colors, and fonts, ensuring seamless integration with their platforms.

“Generating a fully-fledged quiz written by AI only takes a few minutes,” remarks Bill, emphasizing the platform’s efficiency. “Our top priority is engagement and finding innovative ways to attract audiences to the local market,” he adds.

The company firmly believes that local newsrooms can “quizzify” a broader spectrum of content, extending beyond news to encompass politics, sports, entertainment, and more.

“As quiz creation becomes simpler, we anticipate a surge in their popularity because customers enjoy them. They’re fun, engaging, and offer a fresh way to consume news—especially on mobile devices,” Bill Miles says. 

> To read next: How DN leveraged gamification to attract a younger audience, and got almost 35,000 new registered users

3. L’Équipe: Leveraging AI to summarize comments and drive engagement

L’Equipe is a leading daily sports newspaper, TV channel and digital platform.  As their team started to explore how AI could be leveraged across the brand, they wanted to put it to use in improving the user experience for community members. Indeed, L’Équipe has a very strong following of users who not only read articles but interact with the site too. More than 400,000 comments a month for 12 millions unique visitors/month. 

“Our readers love to read comments, and certain articles can generate 1,500 to 2,000 comments”, says Emmanuel Alix, director of digital at L’Equipe. 

Users don’t have to subscribe to leave a comment. They do, however, have to  create a free account to read or add a comment. Once registered, the user enters the top of the conversion funnel and is 15 times more likely to subscribe, says Emmanuel. 

To increase the value of the comments section even further, the team has developed an AI tool to summarize the comments in a drop down box placed within the comments section. Just a single article can generate more than 300 summaries from the comments section.

With this new tool, the L’Equipe team can track how many users click on the “read more” tab to read the full summary. So far, more than 200,000 users have used the feature and the feedback is positive. 

L’Equipe uses OpenAI to generate summaries. The tool isn’t perfect yet, says Emmanuel Alix, but AI does offer certain opportunities for the media.

> To add to your reading list: Changing the way we do paid: how L’Équipe halved subscriber acquisition costs whilst increasing both attribution and last-click subscriptions

4. San Francisco Chronicle, Hearst Media: meet ChowBot, your AI-powered restaurant guide

AI tools are rapidly evolving. How can Hearst newspapers, which include leading local newsrooms such as the San Francisco Chronicle and the Houston Chronicle, and their teams experiment with it? 

Tim O’Rourke, VP of Content Strategy, leads the Dev Hub, an innovative team in Hearst media. His team leads the efforts for the digital innovation for Hearst’s newspapers with data-visualization and interactive projects. 

Naturally, Tim’s team came to experiment with AI. The mindset is to experiment whilst still holding themselves to a high standard and not eroding the public’s trust. “In some areas, the technology is just not good enough yet. We don’t want to be taking risks with our reputation,” Tim explains. 

But they still see huge potential for AI in their work. Production P, for instance, is an internally developed tool to help editors in some of their smaller newsrooms create summaries, and tweak headlines. On the reader-facing side, Tim’s approach is to be very careful. 

Their first AI-powered user-facing product is the “Chowbot”, an AI-powered chatbot that went live in February on the San Francisco Chronicle’s website, helping readers find recommended restaurants and specific dishes in the Bay Area. As the chatbot only uses verified data from the San Francisco Chronicle, the risks of hallucinations and errors are limited. 

Chowbot is reserved for the Chronicale’s subscribers only and was already being used by 10,000 users just a month and a half after its debut.

5. Newsquest: an assistant tool to help local journalists spend more time in their community

When I talked to Jody Doherty-Cove, he was just back from the Nordic AI in Media summit. As Head of Editorial AI for the the second largest media group in the UK, Jody has a lot to share about the impact AI can have for journalists. 

The media group owns more than 250 local titles – including the Berrow’s Worcester Journal, the oldest surviving newspaper in the world. 

“We have AI-assisted reporters using AI tools to help speed up certain tasks”, says Jody. For example, an internal GPT, that writes articles from trusted sources. 

“We only use trusted sources that can be transformed into a media article”, says Jody. In smaller newsrooms, reporters would often have to produce content at scale. But with the help of this tool, they have more time to report on the ground, develop sources and be part of the community they are covering. 

Newsquest papers have seen success thanks to these tools. Subscriptions are up, says Jody, and page views are also doing well. 

“But we also can’t quantify the impact on the editors themselves. They have more free time on their deadline days”, he adds. 

Bringin AI to the newsroom hasn’t been as hard as one can imagine: “It is an easy sell. We found that things those tools are good at are things no one wants to do. We are not taking away things that people enjoy”. 

Beyond the AI-assisted reporters, no reporter or editor is forced to use those tools.

The Audiencers' Festival in London in October 2023 did not disappoint! With 145 digital publishing professionals the afternoon was filled with insightful conversations both on and off stage. To reflect this, we're publishing not just one, but a series of takeaway articles from the event!

Our final session of the day: The art of translating between editorial and commercial

On the panel:
- Maria Bissendorf, Head of Membership Development, now Head of Transformation at Hearst UK
- Morten Ro, Independent engagement designer and advisor

Instead, collaborate and explore a new, shared language. The only place to look for the truth is by connecting deeper with your audience, those at the strategic core of it all.” 

A membership / audience-first pivot essentially looks like this:

The most powerful role in this pivot is a bridge role. This bridge-builder, translator and collaborator will be able to do the ground work between teams before trying to gain support and buy-in from senior leadership, and using patience and experimentation to show the benefits of cross-silo and cross-language efforts. 

And this is the context of Maria’s work at Hearst UK.

3 years ago, the company put their foot on the pedal to launch a digital membership model. But what wasn’t in place back then was a structure and mindset to support this move, and drive success – such as membership or engagement roles, product teams, investment into product, customer experience and UX thinking, etc.

At the time, digital was a scale play, built around traffic and reach rather than bringing audiences back and keeping them, understanding what the reader is doing and why. Afterall, memberships are a product whereby teams need to turn digital content into an engagement play.

Changing organizational charts and KPIs is the simple part of the process. What’s tricky is getting people on board with this new direction and approach, changing mindsets on what’s good“

> Also from The Audiencers’ Festival: Bridge roles in practice: the models, strategies and structures for success

A new mindset that connects old languages at Hearst

Having previously worked in marketing and events, Maria moved into this new thing called ‘membership’ in 2020 with the goal of launching ASAP. However, she quickly realized that the structure wasn’t there. Whilst subscription targets could be hit with little collaboration in a print-driven mindset, siloed teams wouldn’t work for digital membership models. 

So, whilst Maria’s role was to develop the membership product, focussing on the ‘what’ alone wouldn’t work. In order to achieve this goal, silos would need to be broken down and mindsets changed to have engagement and conversion as key performance indicators, instead of reach.

This involved 3 key areas of work: 

• Ways of working: membership will bring the organization together organically, with constant product iteration, paywalling the website and full funnel approach. This means challenging habits and familiarity, and being prepared to fail, something we naturally avoid
• Building a shared language: this often happens organically as you begin collaborating
• Reorganizing: some barriers need to be broken down. The key learning here, for Maria, is that senior leadership needs to be on board. “It’s amazing how fast barriers can get removed if the senior leadership is on board”

Difference in languages: how did that show most concretely?

“The same conversations with different teams had very different outcomes!” 

There’s a huge spectrum between teams: 

• Marketers are often the most chameleon-like
• Editorial and tech are usually the furthest away from each other as their worlds didn’t meet before. Editorial teams are creatives, storytellers, ideas machines, opinion leaders. They’re quick decision-makers but also quick to change their minds
• Tech has their own language (literally!), many acronyms, all about detail, sprints are locked in, juggling lots of dependencies
• ALL teams are flexible and can respond to the user/reader needs, but the challenge can be bringing them on the same page with what we’re doing, or not doing (just yet), as the lead times that they’re used to are incredibly different

How can you bring these two teams closer together? 

• Separated briefing sessions to stay within each teams’ language and give them the details that only they are after (time efficient), then run combined status sessions to build trust and respect
• Build a shared understanding for what we do right now, what we do later, what we don’t do and why, and keep re-enforcing that where needed
• For any ideas that we’re not using right now, save them for later so everyone feels their voice is heard. At the same time, this promotes a focused approach to continue moving in a single direction, which keeps more techy folks happy
• Understand that some words are particularly delicate, mostly the very basics, such as “readership” vs “user” vs “customer” vs “audience”. It’s not about arguing over which to use but making sure everyone’s on the same page and understands each other

The skills needed for this bridge role

Absolute clarity:

What are we doing, why are we doing it. But just as importantly, what are we NOT doing, why are we NOT doing it?

Why is this such a challenge in itself? 

• Leaders and managers aren’t often aware of all the projects happening across the organization, which means they don’t have an accurate perception of people’s workload
• Leaders and managers aren’t aware of the “multiplier effect” — that one initiative in one department could require resources from another, which could stretch them too thin
• People leave or are let go, but their work is simply transferred to another employee

Empathy: 

“I hear this a lot where teams clash and people mistake ‘someone that’s not easy to work with’ with ‘unaligned objectives or difference in language‘.”

Communication: 

Consistently speaking to all teams, which is probably the hardest part because it’s both proactive (e.g. giving KPIs) and reactive (e.g. catching when something goes wrong and correcting it)

“The way to be successful in a bridge role previously came down to your personal relationships, your social capital and your network within the organization,” says Robin Kwong, new formats editor of The Wall Street Journal, on the Journalism.co.uk podcast.

“Now that it’s more mature, it really comes down to having common goals and documentation, and set processes for how you communicate with each other.”

How do you decide what is truly not worthwhile?

Ask your teams, regularly and often, “What are you doing that you don’t think is a good use of your time, and why?”

The American Press Institute recommends this framework:

“In the top-right quadrant are items that both drive revenue and are aligned with your news organization’s mission — enterprise reporting that is shown to engage readers and drive subscriptions, for example. Things that fall into this category of work are likely worth continuing.

In the bottom-left quadrant are items that neither drive revenue nor are central to carrying out your mission. Consider this category the “low-hanging fruit” in your effort to cut out low-value work.

Things that fall into the top-left quadrant — they drive revenue but aren’t “mission-critical” — may be worth continuing and can help “subsidize” the work in the bottom-right quadrant, which doesn’t necessarily drive revenue but nonetheless is central to your mission.”

Tips for staying patient and keeping focused

Reflection:

“Back to my point about empathy, you have to ask why something is not working. What do they need to hear? How can I be honest with them but also give them what they need? For example, discussing what is and isn’t in scope with an editorial team. How we have to start small and simple, something that they’re not used to thanks to their print mindset where they could only send the ‘perfect product’.”

Be comfortable with success not always being defined by KPIs:

“There’s no doubt that we need KPIs, but realizing that you’re making headway in bringing teams together is just as valuable. For example, I don’t yet have an engagement target for our products but there’s nothing more satisfying than having conversations with an editor and a marketeer whereby the editor has an idea, the marketeer underlines it as they can see the customer in front of them and I can imagine what this looks like as, for instance, the app.”

Morten’s perspective

“Building audience and membership mindsets is a marathon, not a sprint, and you need allies in other parts of the organization. Look for bridge roles, creatives, non-ego and low power people who aren’t protective of their own silo or language. Build a creative and playful space to explore where your languages and success metrics overlap and differ (this might be in your spare time if needed). Build a catalog of experiments that you’d love to do. Wait for the right opportunity to pitch. Pitch experiments as a learning effort and document extensively to make learning a success metric in its own right.”

A final takeaway? 

You’re not alone! You’re actually at the center of it all.

Learn the internal politics, but don’t choose a side. You’re the audience advocate.

Senior leadership owns the keys to your success – are they bought in and do they support you and your role? Offer a way to document/illustrate that your discipline is different and important.