> At the FT, subscribers who comment are up to four times more engaged than those who don't, and commenting does more for habit strength than anything else. It is also one of the rarest behaviours on the site

> The fix was deliberately small. Two-thirds of the way through an article, a single short question inviting readers to pause and, if moved, join the discussion below

> Generative AI proposes three questions per article, the editorial community team picks one or edits it, and nothing publishes without a human. The aim is scaling a slow task, not replacing judgement

> The prompt does the heavy lifting: chain-of-thought reasoning about the article first, then rules on tense, abbreviations and opening phrases, shaped by research sessions with the community team

> Comment views rose 3.5%, and 11.5% of readers who hadn't looked at comments in the previous month did. The unplanned win was tone, with fewer off-topic rants once people had something specific to answer

> FT Strategies has run its AI Launchpad with 48 publishers. The backdrop is a discovery crisis: generative search answers the question and removes the click, and referral traffic is eroding

> Aktuality in Slovakia used AI to enter a vertical it had no expertise in, generating travel guides from input variables. Generating section by section rather than whole articles produced noticeably fewer hallucinations

> Il Messaggero built a tool turning raw data into infographics, replacing work it had outsourced. The prompt was the hard part, and models differ markedly in how well they handle visual media

> The FT's Next Best Offer model uses more than 350 inputs to pick which offer a trialist sees at renewal, optimising for lifetime value rather than conversion. A/B tests show significantly higher save rates

> Aliya Itzkowitz's framing is worth borrowing. The AI story for news is usually about scarcity and what is lost. All three of these are about abundance: more topics covered, more formats possible, more tailored offers

> The Financial Times found readers who read comments are 11 times more engaged than those who don't, and commenters 46 times. Yet half of subscribers read comments and only 4% write them

> The FT treats that gap as a design problem, not an audience problem. Its live Q&A feature sends questions privately to journalists rather than publishing them, making the entry point less intimidating

> THE CITY, a New York nonprofit, mails printed postcards carrying explanatory journalism, a QR code and reader questions. Journalists write back to thank people, closing the loop and showing readers they shaped the coverage

> The postcards earn strong evergreen traffic and sponsorship, and sidestep search and AI disruption entirely. They are not, however, a strong driver of immediate donations, and the approach still needs scale

> Ryan Kellett's advice on creator journalists is the least comfortable. Find the staff already at home in those spaces, resource them, and resist the urge to squeeze them into the house brand

Peter Brumby is Product Manager at Financial Times

> The Financial Times built its first North Star around RFV, scoring recency, frequency and volume of reading over 90 days. Above 18.2 a subscriber counts as engaged, and engaged readers cancel 10% less

> That single number changed what got built. myFT, a lightweight way to follow chosen topics, came out of it, and merely becoming a myFT user raised engagement by 86%

> The metric has moved twice since, to lifetime value in 2020 and Global Paying Audience in 2023, targeting three million by 2028. Each built on the last rather than replacing it

> Underneath sits Metrics That Matter, a shared framework for seven departments so every team can see how its work feeds the top number. It also takes pressure off the North Star itself

> A good North Star is single, measurable, time-bound over three to five years, and bold enough to cause slight discomfort. A vanity metric feels good without changing any decision

Instead of being a threat, successfully retaining subscribers can often happen thanks to easy-to-cancel user experiences (as well as retaining throughout the user journey of course). As Travis Bernard, Executive Director, Growth Marketing at LA Times put it, "brands focused on long-term growth prioritize trust over short-term barriers". 

Let's dig into the topic.

Over in France, a recent “termination in 3-clicks” law requires that consumers should always be able to end a contract 1) online and 2) in 3-clicks or less.

Other countries are already implementing similar restrictions, or set to do so soon, and publishers are getting worried…

If someone can unsubscribe in just a click or two, what’s going to happen to my retention rates?!

On the contrary, these regulations could be the push that publishers need to put audience needs first, implement retention strategies throughout the user experience (not just as an afterthought) and allow for positive, lasting impressions that leave the door open to potential re-subscription in the future.

> If you’re interested in learning about how publishers have already reacted to these regulations, in France and elsewhere, check out this article

When should you think about retention?

Or rather, when shouldn’t you think about retention…

Retention shouldn’t just be an afterthought at the moment of cancellation, nor as a ‘post-conversion problem’. In fact, many successful digital reader revenue teams are mission-based to avoid siloed acquisition and retention teams, ensuring the former’s KPIs (high conversion rates but little thought for engagement/retention after this moment) don’t create problems for the latter (high churn rates).

Newsletter subscription, free registration, apps, games, or even simple access to content can all have a positive impact on retention even before a reader subscribes. For instance, increasingly more publishers with a registration model are leading new members through a 3-step onboarding journey, just like they do for subscribers, with the goal of boosting engagement and ticking off some of those important actions that lead to long-term loyalty.

South China Morning Post has a 3-step onboarding journey for newly registered members, encouraging them to sign up to one or more newsletters (habit forming) and build a personalized news feed (support engagement, and this is likely a feature that maximized retention).

The point I’m trying to make is that easy-to-cancel laws are not the nail on the coffin for publishers, because it’s simply this single moment that matters for retention. There are so many other opportunities to keep a reader engaged. And perhaps these easy-to-cancel laws are exactly what’s needed to remind us of this fact.

What’s more, with simple, transparent unsubscription journeys, readers won’t lose trust or confidence in a publisher and may even consider coming back, especially if they know that they could choose to leave again easily without having to phone 5 helplines.

Using data to discover which features correlate with high engagement

In order to think about retention even before a subscriber considers cancelling, you need to understand which features or actions correlate with high engagement, and thus retention.

There are generally two key metrics used to evaluate the value of a feature or product in a reader revenue model:

• How popular a product is (proxied by the average proportion of users that use it at least once over a 30-day timeframe)
• How impactful a product is on engagement (proxied by how engaged a user of that product is 30 days after using it)

DER SPIEGEL did a similar analysis, shared in an article on The Audiencers:

1 – Identifying all available platform features, defined as functional elements that enhance user experience. As news constantly changes, editorial content is excluded.

2 – Analyzing consistently available features and highly engaged users (based on engagement score) to ensure reliable data.

3 – Regular usage means engaging with a feature weekly over four weeks.

They also ran guided in-depth interviews to gain insights into which features users are aware of, which they use and how they rate them. 

According to FT Strategies, you can then rank features by how ‘likely’ they are to drive adoption and generate engagement. Below, an example from a Portuguese publisher.

DER SPIEGEL’s version looked like this:

This research can then be used to strategically nudge those with lower usage towards engagement drivers by highlighting them in mailing, on-site, during onboarding, etc.

The example of Financial Times

At FT, data analysis has shown that the following habits have the greatest impact:

• Signing up to the app & receiving push notifications
• Organically following topics on myFT
• Signing up for newsletters
• Engaging with comments
• Reading on a Saturday or Sunday
• Reading the FT across multiple devices
• Viewing graphics / visual journalism

If we take the example of a subscriber building up their myFT, by following topics and saving content, we can see this being nudged throughout the user journey, from early life, in-life, growth and renewal / cancellation.

In the onboarding journey:

I really like these bundles of content (“newsletters, podcasts and more”) that allow for content diversification around a topic that interests the reader, and all in a single click.

Whilst exploring the website (‘Add to my FT’):

On the subscriber account page:

Readers are congratulated for completing this action (excuse the bad screenshot):

And finally it’s included in the cancellation journey for loss aversion:

Reminder of the goals in the cancellation journey: 

TL;DR: to both retain the subscriber whilst not harming potential future re-subscriptions or referrals.

> Create a fair amount of friction to prompt reconsideration
> Not be too obstructive to cause frustration
> Highlight the value loss
> Understand why they’re leaving to be able to…
> …offer relevant alternatives
> Encourage them to commit for a longer duration
> Ensure they’re confident in the fact that they’ve cancelled but can come back if they want to

For Romain Lhote of L'Équipe, personalized offers, messaging and showcasing relevant usage stats (such as number of articles read) during cancellation proved highly effective.

> More benchmarks on the topic in this dedicated article.

What other strategies should you try to keep readers from visiting the cancellation page?

> Understand the conversion killers and address them

It’s not enough to understand why someone wants to use your product. You also need to know why not?

At Freeletics, they realized early on that many people don’t have much space to train at home. So they developed special workouts that can work in small spaces. Other users have neighbors who complain about the noise, so they need quiet workouts.

Applying this to the media industry, EL PAIS has addressed the “Lack of time” restraint and created an “Exprès” feature, offering readers a curated update on the most important news stories of the day in 7 headlines, explained in 7 minutes.

FT Edit was launched with a similar goal, also providing a lower-cost subscription offer.

> Lock-in effects

This refers to the idea of making it a hassle to switch between products or cancel a subscription. But, be warned, it’s a careful balance – if you overdo it, customers will feel trapped and frustrated, thus damaging the brand. Or, in the worst case, you end up creating dark patterns that really aren’t recommended (slash are illegal).

But there are little innovative tricks to keep your customers loyal…

For example: personalization

Spotify is the prime example here. The app knows your music tastes from the last 10 years and uses this data to recommend new music and create tailored playlists. Spotify itself says that its promise is not just access to music, but more importantly discovery. Switching to Apple Music means the app has to get to know me all over again before it can make recommendations on a comparable level.

A second lock-in example: price guarantee

Average monthly spend is increasing rapidly as prices soar. This makes old contracts all the more valuable. This can be the flat lease from the 90s or, as with the Washington Post, a price guarantee for the next 50 years. For the Watergate anniversary, they offered a subscription for $50 per year with a price guarantee until 2072 – which, importantly, expires if you cancel or pause the subscription.

> The brilliant Lennart Schneider shares 9 lock-in effects in his article on The Audiencers

> Habit forming

Subscription models live off habits. Just look at the success of newsletters in retaining subscribers, or the buzz around Wordle, returning to complete the game on a daily basis.

But hardly anyone is as good at forming habits as Duolingo.

A large part of this success is thanks to streaks, a feature where users are rewarded for completing an exercise a day.

But this strategy wasn’t built overnight – the team have run over 600 tests on optimizing streaks. In Lenny Rachitsky‘s podcast, Jackson Shuttleworth (Group PM, Retention Team) reveals what they’ve learned through testing:

• Find a goal that people want to achieve

Streaks don’t work for every digital offering. They’re designed to build habits that users are proud of and that are difficult to achieve.

E.g. at beehiiv: “I would like to write a newsletter every week”

• Remember that streaks rely on two psychological effects:

> Loss aversion: psychologically, it’s worse for people to lose something than never to have had it.
> Sunk costs: the more time, money or work we’ve already invested, the more likely we are to hold on to things

That’s why the perceived value of your Streak increases with each passing day. And so does the effect on retention.

• Find the right balance between ambition and flexibility

Jackson sees a great danger in streaks becoming too easily accessible and therefore worthless. At the same time, retention drops abruptly if a streak is lost.

Hence why they introduced “streak freezes”. If you take a day off, you can still continue your streak.

> Have a flick through my LinkedIn post on the topic for more recommendations from Lenny.

> And if all else fails, you always have the grace period and post-cancellation

Prioritizing former subscribers in your strategy, especially reaching out to them within the first few days after they cancel, is crucial. At L’Équipe:

• 70% of re-subscribers re-engage within seven days of cancellation.Former subscribers acquired during promotions are particularly responsive to targeted relaunches with attractive pricing.
• On average, former subscribers are 18 times more likely to re-subscribe than members, while members are 18 times more likely to subscribe than anonymous users.

It’s for this reason that they target subscribers during their grace period and immediately post-loss of access.

For instance, a banner warns churned users that they soon won’t have access to this content:

Or a pop-up with a message from the Editor-in-chief sharing what’s coming up at L’Équipe:

Article written by Nicolas Galland, with Loïc Jalmin. Thanks to Madeleine White and Maria Frih for their fantastic reviews and feedback.