How Alma Media and Mather Are Using AI to Rewire the Economics of News Subscriptions
Why Alma Media’s AI paywall experiment matters for subscriptions and data: early results show how algorithms can boost growth, preserve advertising reach, and redefine the value of quality journalism.
The Finnish experiment
Alma Media and Mather Economics reveal how AI-driven dynamic paywalls are reshaping subscriptions, editorial strategy, and the business of quality news.
In 2012, Alma Media became the first publisher in Finland to launch digital subscriptions. The timing looked ambitious. Paywalls were spreading across Europe, but the model was unproven outside the English-speaking world. The early years were slow. Metered access a common choice at the time proved ill-suited to the company’s readership. Consumers showed interest, but growth stalled.
By 2017, the firm had rethought its approach. Out went the metered wall; in came a premium model. The change was more than technical. For the first time, sales targets were set for journalists. Editorial staff shared key performance indicators with the sales team. The result was closer collaboration across the newsroom and the commercial side a shift that many publishers still struggle to manage. Growth quickened.
Then came the pandemic. Like elsewhere, consumer subscriptions spiked. Traffic surged, and readers paid for reliable information. But the COVID boost soon levelled off, leaving Alma facing the same question as its peers: where would the next wave of growth come from?
Johanna Suhonen, Alma’s VP of Content Business presents at Press Gazette’s Future of Media Technology Conference in London on 11th September 2025. Image © Press Gazette.
Shifting to data-driven paywalls
Alma’s answer was to bet early on artificial intelligence. An internal AI team was created two years ago. The results were modest at first: about 2% of subscription sales were directly attributable to AI-driven interventions. But the direction was set.
The breakthrough came in 2022, when Alma partnered with Mather Economics and adopted Sophi, a dynamic paywall system developed with The Globe and The Mail in Canada. The tool uses AI to judge the subscription potential of individual articles and the likelihood that a reader will convert at a given moment. Unlike a static rule-based wall, it adapts in real time, weighing advertising revenue against subscription opportunities.
Early tests were striking. Subscriptions rose by 37%. Registrations the critical first step towards first-party data capture jumped by 130%. Log-ins increased too. Perhaps most importantly, intercept levels, the points at which readers hit a wall, stayed within the agreed 2–5% range. That preserved advertising reach while generating new subscribers.
Sales targets in the newsroom
One of the more unusual features of Alma Media’s model is that journalists are given subscription sales targets. In most news organisations, commercial objectives are kept separate from editorial work. At Alma, the link is explicit.
Reporters are expected to produce material that not only informs but also persuades readers to subscribe. The aim is not to chase clicks or volume, but to encourage journalism that demonstrates measurable value. A carefully reported article that convinces a small but influential group to pay is considered more valuable than a widely read story that produces no conversions.
The practice remains rare in publishing. Most companies avoid it for fear of eroding editorial standards. Alma argues the opposite: that clarity about objectives strengthens the newsroom. Journalists see directly how their work supports the business, and management gains a clearer view of which kinds of reporting underpin growth.
The effect has been a shift in newsroom priorities. Editors now weigh stories not just by reach or visibility, but by their ability to drive subscriptions. Over time, this has begun to reshape commissioning and placement decisions, aligning day-to-day journalism more closely with long-term sustainability.
Editorial scepticism, then relief
For any newsroom, the introduction of AI raises uncomfortable questions. Would algorithms overrule editorial judgement? Would business logic trump independence?
Johanna Suhonen, Alma’s VP of Content Business, admits scepticism was widespread at first. Editors wondered how their workflow would change, or whether an algorithm could possibly understand the nuances of Finnish content. But once results came through, doubts gave way to relief. Tasks became easier. The system identified stories with hidden subscription potential that journalists had overlooked. Workflows were streamlined, and the outcomes spoke for themselves.
“The editorial team was happy,” says Suhonen. “They had fewer guesses to make, and better results to show for their effort.”
The balancing act: ads versus subs
If editorial independence is one challenge, advertising is another. For decades, the industry’s growth logic rested on maximising reach. Subscription models upend that. Every wall risks reducing inventory and eroding ad impressions.
Alma tackled the tension by setting clear rules at the outset. They agreed that the proportion of readers stopped by the paywall, the intercept rate, could rise or fall within 10% of the baseline. This safeguard allowed testing to continue without putting advertising reach or reader access at risk. Certain traffic sources remained free to preserve reach. Above all, the company relied on data rather than assurances. Over time, the advertising team saw that revenue could be protected while subscriptions grew.
This balance where AI learns to place friction in the right spots is becoming the core of sustainable digital publishing. A dynamic pay wall does not treat every reader alike. The occasional visitor may be left alone, preserving ad yield. A registered user showing repeated engagement may be prompted to subscribe. It is a system tuned for parallel goals rather than single-minded pursuit of volume.
Personalisation and the value of registration
For Mather’s Ariel Burkett, VP of Global Sales and Marketing, the emphasis now is not just on selling subscriptions but on building a base of registered users. Registration is the gateway to first-party data the asset that underpins both advertising and subscription models in an era of tightening privacy rules.
That dual-track approach makes registered users central. They generate the behavioural signals that feed AI, allow for personalised offers, and protect publishers against the decline of third-party cookies. In effect, registration is the hinge between audience scale and customer value.
Ariel Burkett, VP of Global Sales and Marketing, Mather Economics at Press Gazette’s Future of Media Technology Conference in London on 11th September 2025. Image © Press Gazette.
“The ability to train different models one for driving registrations, another for driving subscriptions is crucial,” You don’t want to sacrifice one objective for the other.”
Ariel Burkett, VP of Global Sales and Marketing
Data headaches and organisational change
Implementing dynamic paywalls is not just about software. Data projects are complex. Dashboards multiply, signals are messy, and the link between editorial practice and conversion outcomes is not always clear. Alma found the technical integration relatively smooth. The bigger challenge lay in aligning the data project with newsroom realities.
Here again, targets played a role. Instead of measuring output in terms of “enough premium content on the site”, journalists were told how many subscriptions their work needed to deliver each week. That changed behaviour. The content strategy did not shift overnight, but editors began to see stories differently as assets with distinct audience values. Some articles mattered hugely to a small readership segment. Others had broad reach but limited conversion potential. Over time, these insights are likely to reshape commissioning, placement, and promotion.
Lessons for publishers elsewhere
Several lessons emerge from Alma’s experiment:
1. Static models run out of road. Metered paywalls may once have worked, but growth eventually stalls. A premium model, tied to quality content and targeted interventions, delivers more.
2. Newsroom alignment matters. Shared KPIs between editorial and sales accelerate growth. Without them, cultural resistance persists.
3. AI can protect independence. Far from undermining journalists, AI can relieve them of guesswork and uncover subscription opportunities invisible to human judgement.
4. Advertising and subscriptions can co-exist. With data and tolerance limits, publishers can grow subs without cannibalising ad revenue.
5. Registration is the new currency. First-party data underpins both business models. Getting readers to sign in is now as important as getting them to click.
Charlotte Tobit, UK Editor, Press Gazette,Johanna Suhonen, Alma’s VP of Content Business and Ariel Burkett, VP of Global Sales and Marketing, Mather Economics at Press Gazette’s Future of Media Technology Conference in London on 11th September 2025. Image © Press Gazette
Risks and open questions
Despite the positive results, challenges remain. Growth curves can flatten again. The novelty of AI-driven paywalls may fade. Readers may tire of prompts if personalisation feels intrusive. And the economics are not uniform. What works for Alma in Finland may not translate seamlessly to larger, more fragmented markets.
There are also strategic questions. Should publishers build their own AI models or rely on external partners? Alma opted to buy, but other firms may prefer to own the technology outright. The trade-off is between speed and control.
A new foundation
For Suhonen, the bigger story is not any single test result but the foundation being laid. “We are building a smart label,” she argues shorthand for a sustainable, data-driven, audience-first business. It is an ambition shared across the industry.
Publishers no longer speak of digital as an add-on. It is the business. The goal now is to combine the economics of subscriptions with the reach of advertising, using AI to manage the trade-offs. If successful, this approach could provide the stability that news organisations have lacked for two decades.
The Alma-Mather partnership will not end the debate over AI in newsrooms. But it shows that algorithms, when designed with clear rules and human oversight, can help rebuild the economics of journalism rather than erode them. For an industry long caught between dependence on platforms and the fragility of paywalls, that is no small promise.
Thanks for reading all the way to the end of the article! This post is public so feel free to share it, and if you have not done so already sign up and become a member.