> Svenska Dagbladet went from 98,000 to more than 158,000 digital subscribers in two years, over 60% growth, and hit its full-year 2026 target by August

> One North Star: publisher Lisa Irenius replaced separate goals for reach, logged-in users and social with one, more subscribers. A weekly reach of 1.7 million Swedes against roughly 100,000 subscribers was read as headroom, not a problem

> Content has two jobs. Quick news builds habits and traffic, while "aha" stories convert. During the Swedish election traffic spiked, but many of the best-converting articles weren't political: readers came for the news and subscribed for something else

> Paid by default: most articles publish behind the paywall, so the newsroom decides what to leave open, not what to lock. An algorithm flags trending open articles so they can be closed mid-life

> Around 80% of SvD's audio subscribers are new to the brand, people text had never reached. And content commissioned for impact rather than conversion often brings in the most subscribers

Christoffer’s background helps explain his view. He started out as a local journalist, joined Schibsted in 2008 on the advertising side, and now runs subscriptions for SvD. So he knows the commercial levers, but also how a newsroom actually works.

Schibsted is a matrix organisation, with separate subscription and advertising organisations. Christoffer sits on the subscription side, responsible for SvD, and his team has worked closely with SvD’s editorial teams for several years.

Because content is the thing that matters most, the subscription team is deliberately clear about where its role starts and stops:

• Make subscribing frictionless. The team runs A/B tests on the paywall to make it as easy as possible for readers to sign up.
• Share data and insight. It gives the newsroom perspective on how stories are performing.
• Leave the editorial calls to editorial. “When it comes to deciding which kind of story to go for, it’s totally up to the editorial team,” Christoffer says.

He describes his team’s contribution as giving perspective and collaborating, rather than steering. For anyone who’s struggled to bring a newsroom along on the subscription journey it’s a useful framing: the business side earns influence by being helpful, not by setting the editorial agenda.

Lesson: if content is what sells subscriptions, the subscription team’s best work is everything around it. Remove friction at the point of conversion, and give journalists insight they can actually use.

One North Star for the whole organisation

So how do you get editorial and business to genuinely work together? When I asked what had helped most, Christoffer didn’t hesitate: a single, shared goal.

When Lisa Irenius became SvD’s publisher two or three years ago, she set a single North Star: reach more subscribers. Before that, SvD had separate strategies for reach, for logged-in users, for social media engagement and more.

It sounds simple, but bringing goals together under a single KPI made it much simpler to understand what everyone was aiming for. It also gave teams a shared starting point for the harder conversations: what do we need to do to get there, what are our strengths together, and what do we need to work on?

The numbers made the goal feel obvious. SvD reaches 1.7 million Swedes every week. When this push began, it had around 100,000 subscribers.

For Christoffer, that ratio between reach and paying readers wasn’t a problem. It was a sign of huge potential.

Lesson: if your teams are pulling in different directions, cut the number of goals. A single, easy-to-say target gives editorial and commercial a shared language, and comparing reach with subscribers shows everyone how much room there is to grow.

Habit content and “aha” content: you need both

SvD is very focused on conversion. But the most useful thing Christoffer shared was how the newsroom thinks about two different jobs content has to do.

Job one: habit. Quick overviews of what’s going on in the world and what you need to know today. This content “doesn’t give us new subscribers,” he says, but it delivers the daily engagement and routines SvD relies on.

Job two: conversion. Lisa talks internally about creating “aha” moments: pieces that offer insight, appeal to the intellect, make readers feel a bit smarter, or simply bring some joy. These deeper stories are what bring in new subscribers.

Neither works on its own. “You need to have both of those to have a news experience that is worth paying for over time,” Christoffer says. One type of content explicitly feeds the other.

Interestingly, SvD doesn’t formally work with the classic user needs model, though Christoffer says it has drawn some inspiration from it.

The big story isn’t always the one that converts

Sweden’s recent election offered a clear example. Traffic spiked, as you’d expect. But the articles converting readers weren’t mainly about politics.

People came for the news of the moment, then found something else on the site worth paying for. That’s why a broad news experience matters: the story that brings someone in isn’t necessarily the one that makes them subscribe.

Lesson: map your output to the job each piece is doing. Measure habit content on engagement and return visits, and treat traffic peaks as a doorway to the rest of your journalism.

Impact first, conversions second

One of SvD’s standout pieces of journalism this year was “Why Did Max Die?” (“Varför dog Max?”), a three-part investigative podcast with accompanying articles. It looked at the death of a 13-year-old boy who had been bullied, how the school system failed to support him, and his mother’s fight for answers. According to SvD, it has led more than 4,000 readers to subscribe.

Yet Christoffer is clear the team never started from conversion. “In that case, we didn’t see it as, ‘Okay, is this going to give us a lot of subscribers?'” he says. The starting point was a sensitive, important investigation that needed to be told, and the question was how it could have the most impact. SvD then scaled it across the formats it does best.

Some pieces are planned with conversion in mind, he adds. This one wasn’t, and that’s exactly the point he was making earlier about content coming first.

Lesson: track conversions closely, but don’t let them drive every editorial decision. Serious, well-told journalism with real impact can turn out to be your best acquisition content.

Paid by default, with data deciding the exceptions

The paywall isn’t reserved for a special premium section. “Most articles are behind a paywall,” Christoffer explains. In practice, the newsroom publishes paid content from the start.

That flips the usual question. Rather than deciding what to lock, SvD decides what to leave open, and when. An algorithm advises the team rather than deciding for it: if an open article starts trending on social media or drawing heavy traffic, it may be moved behind the paywall along the way.

The next step is a dynamic paywall, setting access levels automatically based on signals such as engagement with the article, where the reader came from and their user profile. The aim, as Christoffer puts it, is to set access levels “without being a part of the process on a daily basis.”

Lesson: if most of your journalism is paid, put your energy into the exceptions, and let data flag when an open article has earned its place behind the paywall.

Audio reaches people text never did

SvD has been producing audio for several years and is honest that it’s still learning. Text remains its core strength. But audio is doing a different job.

Around 80% of the people who subscribe for audio are new to SvD, Christoffer says. These are readers, or rather listeners, that SvD had never managed to reach through text.

That raises the question SvD hasn’t fully answered yet: retention. How do you keep someone who arrived for a podcast? Christoffer sees two routes. Help them discover text journalism, and build other things they’ll want to stay for.

Getting more out of the text you already have

SvD hasn’t tested AI-generated article summaries. Instead, it has repackaged its journalism in three ways:

1. Read-aloud articles. Rather than AI voices, a reporter reads the article. This has had some success.
2. The Morning Brief newsletter. Sent at 7am every day, it covers what happened while you were sleeping. Its value is habit: it gives readers a reason to start the day with SvD.
3. An AI book recommendation bot. Built on the culture section’s archive of book reviews, it drew strong engagement, though Christoffer stresses it’s early days.

Across Schibsted, most AI work so far has gone into tools for journalists rather than reader-facing products.

Lesson: treat new formats as a way to reach new audiences, then plan the bridge to the rest of your product. And don’t overlook your archive: it can be the raw material for genuinely useful reader tools.

Free for 18 to 25 year olds: lowering the barrier

SvD now offers full free access to 18 to 25 year olds. It launched ahead of the Swedish election, largely as a marketing campaign, so it’s too early to know how many will convert when they turn 26.

Christoffer is refreshingly realistic about it. Given SvD’s current product, he thinks it would be optimistic to expect many people in that age group to pay right now. SvD is a text-heavy experience, competing with social platforms and content creators for younger audiences’ attention.

So the aim is simpler: remove the barrier, let them in, and build a relationship over time. Just as importantly, it gives SvD a group of younger readers to learn from. What works for them, what doesn’t, and what they’d like to see more of.

For anyone looking for inspiration here, Christoffer points to Amedia in Norway, who ignored older subscribers in dashboards to attract younger ones.

Lesson: a free youth offer is an investment in relationships and insight, not a quick revenue play. Watch how they engage and listen closely, so that the product you build for them later is one they’ll want to pay for.

> Plenty of newsrooms use the language of user needs but never audit against it. Tagging 75 pieces per need by hand is slow, and taggers work from instincts they've never written down

> Let an LLM take the labour. Tag a calibration sample of about 100 headlines yourself, with notes on the hard calls, then have the LLM tag your whole archive against a written rubric

> The bigger win is alignment. Ask the LLM to turn its disagreements with you into rules, such as "a research finding is UNDERSTAND, however flatly the headline states it", then approve each one as a newsroom

> In Khalil A. Cassimally's tests, the LLM agreed with his blind tags around nine times in ten. Re-check 20 random rows after every rubric change: agreement under about 14 out of 20 means a rule is missing

Before any of that can happen, we need to know where we stand. That’s the audit: tagging a body of recent content by the need each piece serves, then putting those tags next to how each piece performed. It’s a key prerequisite for implementing user needs, because every decision that follows depends on it. And it produces the chart that tends to change minds in a newsroom: what we publish set against the value people get from it.

In my experience, plenty of newsrooms use the language of user needs without ever having audited against it, and others did it once, on a sample, and never went back. There are two reasons for that, and AI can help with both.

Why the audit matters

The audit matters, first, because it gives us a baseline. It shows how our content serves each need today, so there’s something to measure any change against later.

It also shows us the gaps. Some needs perform far better on whatever measure we care about than others do but they’re rarely the needs most of our content serves. That mismatch, laid out in one chart, is what I often find persuades editors and leadership to act. It tells us which needs to prioritise, and which to produce less of.

Plus, the tagging itself is how a team learns user needs. In my experience, discussing whether a headline is there to update or to explain teaches the needs better than a training session does. Done together, the audit can leave a newsroom with a shared sense of what each need looks like in its own content.

> You should also sign up for Khalil’s user needs course: implement user needs in your newsroom

The two challenges that block an audit

Two things tend to stall the audit. The first is the sheer labour. Tagging means someone reads each piece, decides which need it serves, and logs that tag alongside the piece’s details and its performance figures. To get a reliable picture, I recommend tagging 75 pieces of content per number of user need. For a newsroom working with eight user needs, that’s 600 pieces. Each one is a judgement call that takes time. It piles up.

The second challenge is alignment: getting everyone who tags to apply each need the same way. If one person does all the tagging, the tags stay consistent, but the work is slow. If a small team shares it, everyone needs training, plus about a month to absorb the definitions.

Both challenges have the same root: experienced taggers work from instinct. Dmitry Shishkin, who championed the model at the BBC World Service, pointed out that it feels familiar to journalists from the start since the user needs correspond to story angles they already use. I’d add that the same familiarity means the rules people tag by rarely get written down. An unwritten rule can’t be checked or taught, and a newsroom can’t explicitly agree on it.

How AI helps

AI can ease both the labour and the alignment problem.

The labour is the obvious one: an LLM can tag an entire archive of headlines in a fraction of the time it would take a person. It also changes the maths. The 75 pieces per user need is the size of the audit. Once the LLM does the tagging, it can take on your whole archive rather than a sample. What you still tag by hand shrinks to a calibration sample of about 100 headlines, used to build and check the rubric the LLM works from.

The alignment help is, I think, the more valuable. When an LLM’s tags differ from yours, you can ask it why, and ask it to turn the differences into rules. It takes the scattered notes you made while tagging, and the calls you made without explaining them, and writes them up as general rules anyone can apply. Once they’re written down, a newsroom can discuss them and agree on them.

Doing this sharpened my own definitions of each user need. Two examples, from the definitions I’d been teaching with to the rules the process produced:

KNOW. 

Before: give me the who-what-when-where; let me know the latest. 

After: the headline states an event in the world and stops there. A research finding isn’t an event, however flatly the headline states it, so it’s UNDERSTAND. A poll of what people think this week is a fact about the world, so it’s KNOW.

DO. 

Before: take action; find useful information for my day-to-day. 

After: the reader has to be the one acting, and the headline has to instruct them. Advice aimed at governments is UNDERSTAND, and so is a question about the reader’s life that weighs the options rather than telling them what to do.

When I ran the rubric on headlines from a publisher, the LLM agreed with my blind tags around nine times in ten: 91% on Claude and 89% on Gemini.

More publishers are looking into using AI this way. Le Nouvel Obs, for one, is exploring AI to classify content though it’s “still very much in the R&D phase”. 

Here’s the process I’d recommend after testing it on 275 headlines from two publishers, all of which I also tagged by hand.

Using AI to audit your content: a seven-step process

Each step exists because the next one depends on it.

The first step gives you a draft rubric, so you aren’t tagging against bare definitions. But a rubric drafted by an AI alone holds its judgement, not yours, so the second step captures yours: you tag a sample and note how you decided on the hard ones. The third step is where the alignment value comes from. The LLM compares its tags with yours and turns the differences into written rules. The fourth step makes sure every one of those rules is your newsroom’s decision before it’s applied to hundreds of pieces. The fifth applies the rubric to your whole archive, which is where the labour saving comes from. The sixth checks a random sample of the results, mainly to find the rules you still haven’t written, which go back through step four. The seventh turns the tags into the chart.

You only tag by hand in steps two and six: just over 100 headlines to build and check the rubric. The LLM does the rest.

1. Draft a rubric with AI, or start from mine

All you need coming into this step is your archive and the user needs definitions you already work with, however rudimentary at this stage.

A rubric sets out, in writing, how to decide the headlines that sit on the line between two needs. Most archives have them: a research finding stated like news (KNOW or UNDERSTAND?), an event headline with an explanation added after the dash (KNOW or UNDERSTAND?), a feature with an emotive headline that’s really an explainer (FEEL or UNDERSTAND?). The rubric is where you settle those decisions.

Give an LLM a few dozen headlines from your own archive, so the rules fit your content, and ask it for a rubric with three things: a definition of each category, the order to test them in, and a decision rule for each kind of borderline headline, with a worked example for each rule.

If you’d rather not start from scratch, the rubric I ended up with is included in full below. You can paste it as a prompt.

Either way, you now have a draft rubric. The next step captures your own judgement to set against it.

2. Tag a sample yourself, then have the LLM tag it too

The draft rubric holds the LLM’s judgement. This step captures yours, which is what turns a generic rubric into your newsroom’s.

Export your archive from your CMS or analytics tool, if you haven’t already: headlines, sections, dates and the performance metrics you care about (you’ll need those in step seven). From that export, draw around 100 headlines, spread across topics and formats. This is your calibration sample. It’s small enough to tag by hand, and it’s separate from the audit itself, which the LLM will run on the whole archive.

Tag the sample yourself first, without looking at any LLM tags. Wherever the rubric doesn’t settle a headline, write a short note on how you decided. The notes matter as much as the tags, because they’re what the LLM will turn into rules. Then give the LLM the draft rubric and the same 100 headlines, and have it tag them too.

Ideally, several colleagues tag the sample on their own as well. Where they disagree with each other, you’ve found your alignment problem at a size you can work with.

You now have two sets of tags for the same headlines. The next step turns the differences between them into rules.

3. Ask AI to turn your disagreements into rules

With your tags, your notes and the LLM’s tags side by side, this step turns scattered judgement into rules people can see and discuss. Give the LLM all three. Ask it to group the disagreements by cause, and to draft one general rule for each group, with a worked headline that shows the reasoning. Ask it as well to flag anywhere your tags contradict your own notes.

Expect a few kinds of rules. Some will generalise reasoning you’d already written in your notes, one headline at a time, into a rule for any headline. Others will write down calls you made consistently but never explained. The second kind can be the more useful, because it’s knowledge that hadn’t previously been put into words.

What comes out is a list of proposed rules. The next step decides which of them go into the rubric.

4. Decide on every rule

The proposed rules are still the LLM’s drafts, and this step makes them your newsroom’s. Every one needs a human decision: accept it or reject it.

Keep in mind that not every pattern deserves a rule. Some disagreements are genuine house calls, better left for a team to talk through. And even when a rule is written down, people can argue with it, and I’d expect those arguments to be how colleagues come to apply the needs the same way.

The rules you accept go into the rubric, which is now ready to apply at scale.

5. Tag your archive

With a rubric that holds your newsroom’s calls, this step applies it to everything you want to audit. This is where the labour goes. Paste the rubric into the prompt, then give the LLM the headlines from your export (and sections, if you have them). Because the LLM is doing the tagging, you can go well beyond 75 pieces per need if you want to – perhaps to the whole archive. Ask for one tag per headline, a confidence level, and a short reason naming which test in the rubric decided it. The reasons make the next step much quicker.

You don’t need to add many worked examples in the prompt. In my tests, adding them moved agreement by a few points at most, because the rubric already holds the detail.

Also: which LLM you use matters less than you might expect. Claude and Gemini were a few points apart in my tests.

Every piece now has a tag. The next step checks how far to trust them.

6. Check 20 at random, to surface the rules you haven’t written yet

The archive is tagged, but however good the rubric, there will be calls it doesn’t cover yet. This step finds them. Pull 20 tagged rows at random, tag them yourself without looking, then compare. For each disagreement, ask who’s right: did you slip, did the LLM, or is this a call the rubric doesn’t cover? The third kind becomes a new rule.

If you can, run the same 20 through a second LLM. Where both LLMs agree with each other against you, check your own tag first. Where both LLMs make the same mistake, there’s probably a rule missing. As a rule of thumb, under about 14 out of 20 suggests the rubric is missing a rule.

Tip: pick the rows at random, not by the LLM’s confidence flags. Some LLMs mark nearly every row high confidence, including the ones they get wrong.

Any new rules go back through step four. Once the check comes back clean enough, the tags are ready for the chart.

7. Plot what you publish against how it performs

With tags you trust, this step sets what you publish against how it performs. That’s why the export in step two needs your performance metrics as columns next to each headline: pageviews (which is what I used), time spent, conversions, return visits, or whichever measures your newsroom values. For each need, set its share of what you publish against its share of each measure. Look at the median per piece as well as the share, so one exceptional piece can’t inflate a whole category.

That chart is your baseline. The next audit will show you whether editorial changes have moved things in the right direction.

Keeping it running

The rubric is a living document. Every random check can add a rule, so date and number each version, and you’ll always know which rules a batch of tags was made under. Mine has been through four versions so far.

From there, the natural next move is tagging new content at publication, with the same rubric, rather than in a big audit after the fact. The mix can then sit on a dashboard, and you see it shift month by month. A settled rubric should also help new staff and new desks learn how the newsroom applies the needs.

What the AI and I tagged differently, and what came out of it

When I compared my tags with the LLM’s, the differences fell into three groups, and each produced something useful.

• Rules generalised from my own notes.
  • The LLM’s early drafts tagged research findings as KNOW. My notes on those headlines kept giving the same reason for disagreeing, and the LLM turned that reasoning into a rule for any headline: a research finding is UNDERSTAND, however flatly the headline states it.
  • On a piece about how a government could protect people from heatwaves, I’d noted that it would only be DO if the action were the reader’s own. That became the rule that DO needs the reader to be the actor.
  • And I’d kept asking myself whether a headline’s emotional register was strong enough for FEEL. The LLM swapped my test for a better one: FEEL needs the piece itself to be the experience, like a review or a personal essay.
• Calls I’d made without a rule.
  • Some of my tags were consistent, but I’d never said why. I’d tagged “How late is too late for dinner?” UNDERSTAND without a note. The LLM wrote down the rule I’d been applying: DO needs an instruction, not a question about the reader’s life. I’d tagged a poll of public opinion KNOW without comment, and the rule now says a poll of what people think is KNOW, while a study of how the world works is UNDERSTAND.
• My own slips.
  • The LLM also held me to my own rules, flagging places where my tags contradicted my notes. I’d tagged a footballer’s sprained ankle FEEL because it was football rather than event news. The LLM had the injury as KNOW, and it was right. The rubric now says sport and celebrity coverage isn’t automatically FEEL.

The table sets the definitions I’d been teaching with against what the rubric now says:

Category	The definitions I started with	What the rubric now says
KNOW (update me, keep me engaged)	Give me the who-what-when-where. Let me know the latest. Keep me up to date with the latest trends. Catch me up with what my tribe is talking about.	The headline states an event in the world and stops there. Research findings aren’t events. A poll of opinion is. Official warnings and service announcements are KNOW. Sport and celebrity news is KNOW like any other event.
UNDERSTAND (educate me, give me perspective)	Help me learn something. Teach me about the wider world. Broaden my horizons. Challenge my opinions.	The piece explains, analyses or argues, including every research finding. “How it works” and “why it happens” both count. An explainer stays an explainer whatever adjectives it wears. Advice aimed at institutions is UNDERSTAND.
FEEL (inspire me, entertain me)	Inspire me to be the best person I can be. Give me hope. Show me a not-so-serious side of a story. Give me something to smile about.	The piece itself is the experience: a review, a human story, a personal essay, a celebration of a person, something told purely for pleasure. Reviews are FEEL. Arguments about culture are UNDERSTAND.
DO (help me, connect me)	Take action. Find useful information for my day-to-day. Help me become part of something bigger than myself.	The reader is the one who acts, and the headline instructs them. A question about the reader’s life that weighs the options is UNDERSTAND.

What to watch out for

AI doesn’t get this right on its own. The process works because the human and the LLM check each other. A few things are worth watching.

• A new rule can fix some headlines and break others. When I added rules, they fixed a few headlines and pulled others the wrong way. So, re-check a sample every time you change the rubric.
• Don’t defer to the LLM, even when it’s often right. Where two LLMs agreed against me, they were usually right. That makes it tempting to accept every disagreement. Every correction I made went towards the LLMs’ answer, so it’s worth asking each time whether you’ve genuinely changed your mind, or just been outvoted.

The real limitations

Beyond those watch-outs, two limitations are worth being clear about.

• One person’s tags were the ground truth. Every number in this piece measures agreement with me. Yours will measure agreement with whoever tags in step two, which is a good reason for more than one of you to do it.
• Some headlines stay contested. A few serve two needs equally, and a rule may not settle them. Those are for a team to talk through.

Start with 100 of your own headlines

The audit is the foundation of user needs work, and it stalls for two reasons: tagging hundreds of pieces by hand is slow, and the people doing it work from instinct they’ve rarely had to put into words. AI takes on most of the tagging. More usefully, it turns the notes and unexplained calls behind that instinct into written rules, which a newsroom can argue over and agree on. The judgement stays with us, in the tags we make and the rules we approve.

You don’t need a data team or a big project to start. Take the rubric below, draw 100 headlines from your own archive, tag them yourself, and see which rules your newsroom would write differently. That calibration sample is the beginning of a rubric that’s yours. From there, the LLM can tag the whole archive, and you can run the audit again whenever you need to.

***

The rubric in full

The four categories follow the top level of smartocto’s User Needs model 2.0. Paste everything below into your prompt, then adapt the rules to your own newsroom’s calls.

What you are tagging

Every piece of content gets one tag: the need it serves for the person consuming it. Not what it’s about (that’s topic), not what shape it is (that’s format). What it’s for.

Four categories, each covering two of the eight needs in the model:

Tag	Covers	The reader comes to…	The headline typically…
KNOW	update me, keep me engaged	find out what happened, or what everyone’s talking about	states an event in the world: a decision, an announcement, a result, a development, a trending moment
UNDERSTAND	educate me, give me perspective	learn how something works, or make sense of what it means	promises an explanation, a background, an analysis, an argument, a verdict
FEEL	inspire me, entertain me	be moved, uplifted, amused, delighted, taken somewhere	signals the register: a review, a human story, something funny, curious or beautiful
DO	help me, connect me	act, in their own life or with others around a cause	addresses the reader as the one who will act, decide or take part

The tests

Run them in this order. Stop at the first clear yes.

Does the headline address the reader as the one who will act? Tips, how-to, should-you, what-to-do, here’s-how-to, join, sign up – aimed at the reader’s own life or their part in something collective. If so, DO.

Does the headline state an event in the world, and stop there? Something happened, was decided, announced, won, lost, or is trending, with no promise to explain it. If so, KNOW.

Is the piece itself the experience? A review, a human story, a personal essay, a celebration of a person, something funny or told purely for pleasure. Strip it to a neutral summary and the value disappears. If so, FEEL.

Otherwise the piece explains, analyses or argues. How it works, why it happens, what a study found, the background, what it means, what the evidence says, the case for or against. That’s UNDERSTAND.

Decision rules for the cases that cause arguments

A research finding is UNDERSTAND, however flatly the headline states it. KNOW is for events in the world. A study result is something the reader learns (educate me), not something that happened to them. “Collagen supplements can help your skin and joints, large new study finds”: UNDERSTAND. “Horses really can smell fear, new study claims”: UNDERSTAND. It becomes DO only when the headline addresses what the reader does with it: “How eggs can help you come off Wegovy”.

A poll of opinion is KNOW. A study of the world is UNDERSTAND. “Three out of four Americans say AI firms are not doing enough to prevent disaster, Reuters/Ipsos poll finds”: KNOW, because a poll reports what people currently think, which is a fact about the world this week. “Collagen supplements can help your skin and joints, large new study finds”: UNDERSTAND, because a study reports how the world works, which is something the reader learns.

An event with an appended promise takes the tag of the promise. The event is the peg; the clause after the dash is the need. “Ebola outbreak declared a global health emergency”: KNOW. “…– what you need to know”: UNDERSTAND. “Taiwan Travelogue wins the International Booker – a deftly translated tale of food, love and history”: FEEL. “Keir Starmer has resigned – but what could a new PM actually do differently?”: UNDERSTAND.

“How it works” and “why it happens” are both UNDERSTAND. Mechanism and meaning sit in the same category: educate me and give me perspective. “Why sepsis is becoming harder to treat in Europe” and “Why climate action stalls, despite widespread popular support” are both UNDERSTAND. You don’t have to decide which of the two it is.

An explainer stays an explainer, whatever adjectives it wears. “This is why you only breathe out of one nostril at a time”: UNDERSTAND. “How Nasa’s Artemis II mission rediscovered the majesty and mystery of the Moon”: UNDERSTAND, an emotive frame on an explanatory piece. “Why was an Egyptian mummy stuffed with a fragment of Homer’s Iliad?”: UNDERSTAND, a curiosity explained. FEEL needs the piece to be the experience: a review, a story, a celebration of a person, a personal essay, something told purely for pleasure. “Forget flowers: lovers in 19th-century Ireland exchanged hair”: FEEL. “My unsung hero of science: Carolyn Wood Sherif”: FEEL.

Reviews are FEEL. Arguments about culture are UNDERSTAND. “Normal: this quirky Bob Odenkirk caper isDie Hard meets Fargo”: FEEL. “Horror won big at the Oscars – it’s time the genre was taken seriously”: UNDERSTAND.

A trending peg with an explanation is UNDERSTAND. The trend itself is KNOW. “Kim Kardashian’s new caffeine-free energy drink – here’s what the science says”: UNDERSTAND. “Prescriptions for ivermectin soar after Mel Gibson claims it cured friends of cancer”: KNOW, because the conversation is the story.

“What you need to know” is UNDERSTAND, unless the need is a personal decision. “Osteopenia: loss of bone mineral density affects millions – here’s what you need to know”: UNDERSTAND. “Vitamin D: everything you need to know – from when to take it, to how much you really need”: DO.

DO needs an instruction, not a question about the reader’s life. “Eight ways to sleep well in hot weather”: DO. “How late is too late for dinner? Does eating earlier help with weight loss?”: UNDERSTAND, because it asks and explains rather than telling you what to do. “US or UCITS ETFs: which is better for Singapore investors?”: UNDERSTAND, a comparison the reader may act on, but the piece weighs rather than instructs.

An official warning or a service announcement is KNOW. “Bus passengers face fines of up to S$500 for playing loud audio”: KNOW. It reports that something has been said or made available, even when the reader could act on it. DO needs the headline itself to instruct the reader.

DO also needs the reader to be the actor. Advice aimed at institutions is UNDERSTAND. “Eight ways to sleep well in hot weather”: DO. “Heatwaves are now everyday disasters – here’s how the UK government can better protect people”: UNDERSTAND, because it is a case for what someone else should do. “What your local council can actually do to tackle the climate crisis”: DO, low confidence, because the reader has some agency there, which is the connect-me edge of the category.

Topic never decides. Health can be any of the four. Politics can be any of the four. Sport and celebrity coverage is not automatically FEEL: an injury, a squad change or an award is an event, and tags KNOW like any other. If you find yourself tagging by section, stop.

Tag the promise the headline makes, not the piece you imagine behind it. When you only have the title, tag the title. When you have the standfirst too, let it settle the call, not overrule it.

One tag per piece. Most pieces serve more than one need. Tag the one that, if unmet, would make the reader feel the piece failed. If two are genuinely equal, tag the one higher in the test order (DO, then KNOW, then FEEL, then UNDERSTAND) and mark the row low confidence.

Output

For each row: id, tag, confidence, reason

tag is exactly one of KNOW, UNDERSTAND, FEEL, DO

confidence is high or low. Low means a second reader could reasonably tag it differently. Low-confidence rows are the ones worth checking by hand

reason is one short clause naming which test fired – e.g. “addresses reader’s action”, “event stated”, “review”, “appended promise: explains”, “study finding – educate me”

> Article for article, "update me" pieces underperform every other user need, yet newsrooms produce more of them than anything else

> Tag every piece with three things: the subtopic (personal finance, not just finance), the format (Q&A, feature, listicle, data-led) and the single user need it serves. Then commission on all three

> One dominant need per piece. An article that tries to serve every need gets a headline that can only sell one, so most readers never discover the rest

> Don't let the newsroom write its own user needs model: take the needs from audience research. At Ringier, across five countries, conversions, subscriptions and loyalty metrics all increased once teams moved to the model

Shishkin’s opening slide was the same one he showed a decade ago, and he made no apology for it. A red line tracks the number of articles a publication produces under each user need; a blue bar tracks the average effectiveness of a single article in that category. The red line peaks at “update me”. The blue bar for “update me” is the lowest on the chart. “Educate me”, “give me perspective”, “inspire me” and “divert me” all do better per piece, and the newsroom produces far fewer of them.

The point, he said, is that publishers have always measured stories, and occasionally formats, but almost never the angle of the storytelling. Once you make the angle measurable, you can make editorial choices about it. “It’s not about what we’re going to cover. It’s how specifically we’re going to cover the story.”

He said the pattern has held in every piece of research since, and the gains aren’t confined to engagement. At Ringier Media International, where he oversaw five countries, every team moved to the model, and conversions, subscriptions and loyalty all went up.

> To read next: How the user needs model increases engagement

The everything article is the weakest article

A natural reaction to the chart is to write pieces that satisfy several needs at once. Shishkin argued the opposite. A piece that tries to update, explain, inspire, and help in one go might seem comprehensive to the editor, but its headline can only promise one of those things, so most potential readers never discover the rest. His rule is that every piece should be dominated by one need. Content built this way, he said, consistently outperforms the general piece that has “a lot of user needs in mind, but not one is dominant”.

The generic model has four areas (know, understand, feel and do) and eight needs within them. Specialist publishers should not stop there. Sarah Marshall built a fashion version at Vogue; Shishkin built a travel model at Culture Trip; ten months ago he and the analytics firm smartocto published the first sports model. Retail, finance and health, he suggested, are still open. Needs such as “help me advance my career” or “help me find the right product” belong in the publication model that serves them.

Three tags, and a different commissioning meeting

For publishers who want something to implement, Shishkin reduced structured data to three fields. First, the subtopic: “finance” is not enough; the tag has to say personal finance or corporate finance. Second, the format, which he was emphatic is not audio, video or text. Those are content types. Formats are the roughly ten shapes a story can take regardless of medium: Q&A, feature, listicle, data-led piece and so on. Third, the user need.

The commissioning conversation then changes. Using a motoring publisher as a worked example, he listed six ways to cover one car launch: explain it to me, help me decide, protect me, price me, show me how, update me. A publisher might concede “help me decide” to a rival and concentrate on “show me how”.

You can use the same tagging to test a mission statement. Most Chief Executives say their publication exists to explain the world, so the share of output tagged “explain to me” should be high. “If your journalism is only 20% explain to me, then your mission statement is wrong.”

Who owns the model

The most pointed line of the session was that newsrooms should be prohibited from writing their own user needs model. A newsroom that sets its own needs and then maps its own output to them is grading its own work. The needs have to come from audience research.

That matters commercially because the model is, in his phrase, the shortest way to unite an organisation. When product people and journalists call the same thing by different names, each quietly stops believing the other. At Culture Trip, his first act was to put senior editors and product managers in one group and get them to agree on the names of things.

Once the vocabulary is shared, product can act on it. If every vertical is producing “inspire me” content, a product manager can bundle it across sport, news and food into one proposition and sell it differently. The harder prerequisite is normalising data across the CMS, email, CRM and video tools; without a single source of truth, any editorial argument can be rebutted with a different dashboard within five minutes.

Beyond content

Two closing arguments bear on retention. First, “commodity content”, the material everyone has and nobody enjoys making, should be automated so that human time goes on original work; most publishers have that balance the wrong way round. Second, content alone will not sustain a relationship. “You need to be helpful to people.” His example was a two-person cycling title that maps Tour de France climbs so readers can plan their own rides. Every niche, he argued, needs a utility like that.

He expects AI agents to do the retrospective analysis, scanning a year of output for underserved combinations of need, subtopic, and format.

How to try it

Shishkin’s advice is to start small: pick one title, section or subsection, and work through four steps.

• Ask what needs your audience has. Take them from audience research, not from the newsroom.
• Check whether your content reflects those needs.
• Check whether your taxonomy captures it. A tag that says only “finance” cannot show which kind of finance story is working.
• Analyse your existing content retrospectively against those needs, then start optimising.

For a large portfolio, the prize is cross-vertical bundling that product teams cannot build today. For a single specialist title, it is a defensible position on one or two needs a competitor has left open. His closing question applies to both: if you disappeared tomorrow, what would change in your audience’s life?

> Metrics look objective. They are not. Every metric carries the business interests of whoever designed it, so a newsroom running on off-the-shelf analytics is optimising for someone else's goals

> Platform analytics are built to maximise ad inventory and time on network. If you sell subscriptions, you are measuring success with a tool designed for a different business model

> The free analytics smaller outlets use are modelled on what the biggest outlets negotiated with platforms. The advantage compounds: winner-take-all monetisation, hierarchies that are hard to challenge, and incentives to consolidate

> Chasing volume shows up in the copy. Articles translated into several languages, wire reports pasted in, headlines bent towards trending keywords. In the AI era the suggestions are no longer keywords but whole sentences, which risks framing social issues more stereotypically

> Goodhart's law: once a metric becomes a target, it stops being a good metric. The argument is not to abandon analytics, but to ask of every number on your dashboard who designed it and what behaviour it rewards

Based on the PhD research of Sophie Chauvet.

What I have found, in a nutshell, is that the bigger the size, the bigger the wins. But to find out about this simplified insight, I first had to take a step back from newsrooms, find the right tools, and ask key questions in order to open the black box of analytics the best I could.  

First, where do metrics come from? A genealogical approach proved to be deeply informative.  

Second, who profits from metrics? Following the money, and identifying who defines them, helped identify the main technical and political dynamics at their core. That’s where the macro approach became fruitful. 

Third, how realistic are metrics? There, I investigated what happens when a quantified measurement captures a social process, and what gets erased.  

After analysing hundreds of white papers, events, and interviews with practitioners across the data monetisation ecosystem, including ad tech, platforms, measurement and analytics companies, there was a single metaphor that helped to understand the stakes of metrics, and why they may negatively re-shape media landscapes: confusing the map for the territory, or how metrics tend to be self-fulfilling prophecies, thus impacting reality based on biased premises. 

The crux of metrics: Self-fulfilling prophecies, or why we confuse the map for the territory

Confusing the map for the territory is an expression borrowed from Alfred Korzybski, which is often used in quantification studies to describe the nefarious impacts of ill-used measurement. It means thinking the measurement model associated with an object is as real as the object. In the context of metrics, this helps to understand why there might be some confusion as to whether clicks mainly represent the audience’s interest, when in fact, they also represent the interests of those who built the model of metrics.  

Let’s break this down further, by answering the first question: where do metrics come from? Here, the history of audience measurement, as told among others by Cécile Méadel, Philip Napoli, and Josiane Jouët, is particularly enlightening. To practitioners in companies such as Médiamétrie and Nielsen, who historically developed studies on audience measurement on television and radio thanks to their Audimat and Peoplemeter, the audience was understood to be more of a convention than a reality, a means to gauge advertising revenue, with measurements mostly agreed upon by the different media involved.  

As such, measurements hold a kernel of truth, but they also serve a function of coordination between players in order to synchronise, predict and project financial benefits, with measurement containing an added layer of filters based on the parties’ main interests, which are also audited in order to ensure some degree of fairness within the ecosystem.  

With the rise of the internet in the late 90s, the first analytics companies, such as Urchin in 1998 that was acquired by Google, emerged with the promise of more realistic measurement. The internet was sold at the time as a new medium where clicks served as the new proxies for direct and real-time feedback from the public, through personal computers.  

This promised to help bypass the often heavy, costly, and slow studies of more traditional audience measurement. This promise was then thought to be fulfilled exponentially with the rise of social media and their troves of personal data waiting to be collected at a scale never seen before.  

This helps explain why the public is sometimes conflated with analytics, or why the map is confused for the territory, as analytics tools entered the newsroom in the 2010s, a field traditional measurement companies hardly ever dared to enter so as to protect journalistic integrity from advertising’s interests.  

Who profits from confusing metrics? 

The rise of analytics in newsrooms coincided with the rise of platforms and real-time data collection at scale. Here, the answer to the second question, who profits from metrics, is answered. Indeed, one aspect that contributed to making platforms so powerful is that their models rely partially not only on providing real time data on website use, but also on collecting this data back to improve their own models, by maintaining a more global overview on all the websites they are implemented on.  

With this broader perspective on audience flows, they are able to optimise advertising models which are key to their financial monopoly over the internet. These models however are black-boxed, as platforms are rarely audited.  

This has severe consequences on the internet ecosystem. For example, recent trials have proved that Google have abused this position of power over data flows by choosing specific, opaque algorithms that prioritised them over news outlets in an undue fashion, leading the latter to lose millions in revenue.  

But beyond the potential for fraud, these black boxes, this network of metrics used by many newsrooms, also contain the different ties of coordination between a slew of internet players that monetise data across the ecosystem.  

This, added to the complexity of the technical layers that sustain online consumption, and the un-audited incentives for inflating numbers, not to mention spam and bots, contribute to a phenomenon of entropy, which help explain why metrics do not always make sense. Disclosing this entropy is not in the interest of platforms. 

Overall in that regard, metrics, when provided by bigger, non-journalistic players, could be regarded not as direct public feedback, but rather as a reflection of bundled interests of the technical infrastructure of the internet, and the political economic negotiations that happen often behind closed doors. 

Metrics have therefore kept their function of coordinating the interests of the players involved, but without much transparency, and with an added layer of realism which is heavily advertised by many internet players, whether in online advertising, or via analytics companies.  

The benefits of scale: Supercharging the Goodhart’s law  

Now there is perhaps one factor that is the most significant when it comes to further understanding why metrics are confused for reality, and why they tend to become self-fulfilling prophecies with deep impacts on media landscapes: scale. This is due both to the fact that metrics are made of statistics, and because online audience flows are dominated by platforms.  

First, the statistical nature of online data flows highlights the importance of scale in monetising data online. This is caused by the fact that most predictions online are based on probabilities, and their associated deterministic patterns.  

This statistical technique helps to observe long term audience consumption patterns and how to optimise the most fine-tuned, cost-efficient audience trajectories according to a specific click architecture. In order to fulfil the promise that predictions become true, scale is a sine qua non condition in order to decipher what historical patterns are the most profitable.  

A corollary of this is that this deterministic parameter tends to cement historical repetitions, and to further benefit media players who existed first, as both benefits of scale tend to generate cumulative effects, and to provide a reputation for reliable profit which is highly valuable when it comes to organising and predicting budgets.  

This one main parameter is what allows platforms, whose main objective is to accumulate data, to maintain a stronghold on media landscapes, by adopting this meta-deterministic perspective on data patterns, and therefore, holding a monopolistic, black-boxed overview of what performs best online. 

By using probabilities and providing metrics as a map, platforms may supercharge the effect of self fulfilling prophecies online, through metrics that often tend to become objectives rather than simple indicators. 

This mechanism is called the Goodhart law. When metrics are used as objectives, they cease to be good metrics. In the case of metrics in news, it means that using scale in order to fulfil the deterministic parameter of finding truth through statistics tends to create self-fulfilling prophecies. By incentivising scale, volume, clicks, and page views, there are some negative macro effects that come as a consequence of mis understanding the power of metrics.  

Macro impact #1: The winner-take-all effect  

First, there is a winner-take all effect. The bigger the media, in terms of volume of clicks and data, the bigger the possibilities of monetising the data fruitfully. This winner-take-all effect is similar to the one observed in the business model of platforms, where scale advantages tend to become cemented and supercharged once a certain volume is reached. The bigger the media in terms of data volume, whether through subscriptions or through the accumulation of page views, the bigger the monetisation opportunities. The benefits of owning historical data tend to be cumulative as the predictions are more reliable and more precise.  

Furthermore, metrics models provided by platforms are often based on those bigger media outlets, as platforms are naturally more interested in those large pools of data that benefit their own meta-deterministic models of data analysis. In that sense, adopting the metrics provided by platforms tends to mean, that one’s strategy becomes symmetrical to platforms’ interests.  

Macro impact #2: Media hierarchies  

A second macro effect of this model is that it tends to also favour media hierarchies that are difficult to challenge. Bigger outlets remain at the top in terms of traffic and monetisation, because platforms tend to model their metrics and software according to those levels of scale and volume, which is what is of interest to them. Bigger outlets are also those who benefit more from in-person negotiations with platforms as to what metrics matter, and those models tend to be those that are then provided in free versions of analytics software that are more accessible to smaller outlets.  

Furthermore bigger outlets with more financial resources are more able to tweak analytics parameters, whereas smaller outlets tend to have more ready-made, black-boxed tools that may not be that adaptable to their own needs. Those outlets who have higher volumes of data also tend to be favoured in negotiations with other players of the data monetisation ecosystem, such as ad tech companies, as they benefit from a solid monetisation reputation.  

This model is not necessarily new, as traditional measurement companies also used similar hierarchical, rankings-based models, where media outlets with more resources and established historical deterministic patterns, have access to more precise and efficient measurement. 

In this digital analytics model therefore, new players have fewer chances to compete in terms of data monetisation because they lack the historical reputation and the reliability of accumulated historical patterns that unlock trust in predictions and funding opportunities. Therefore, the parameters of determinism, scale, and self-fulfilling prophecies are revealed by the simple fact that measurement and metrics are based on statistics and projections.  

Macro impact #3: Media concentration  

The data monetisation benefits provided by scale include a third significant macro impact on media landscape: media concentration incentives. In order to compete in this quantified, rankings-based system, some media outlets have adopted strategies to access those scale benefits. By creating alliances, accumulating media outlets, and re-joining databases, more volume can be reached. In that sense, media conglomerates can be incentivised to optimise online advertising and data monetisation opportunities, with a potential risk for media diversity.  

Macro impact #4: Devaluing journalism 

The final macro consequence that derives from the volume incentive, which is perhaps the most significant of all, is the devaluation of journalistic standards. From a strictly volume-oriented data monetisation perspective, it makes sense to produce more content which translates into more clicks, at the expense of journalistic standards. While clicks and webpages can easily be inflated from a quantitative perspective, the labour that tends to come with quality, original reporting, which often takes time, tends to be disregarded, from the perspective of this quantified financial model.  

In that sense, favouring financial value to access scale benefits tends to empty journalism from its values, and the value of its labour. Some strategies adopted by some media outlets include quickly producing articles at scale, by translating them into different languages, copy-pasting news wire reports, and coordinating with trending online keywords that do not necessarily stem from organic audience behaviours or from journalists’ flair.  

Journalists have sometimes conceded, when interviewed for this research, that they felt they had to twist their instincts in order to ensure content production that may have felt industrial rather than the byproduct of careful listening of their public’s needs. On the brighter side, another strategy also includes the attempt to quantify the value of journalism’s democratic mission, by promoting the quality of the involvement with the public as a selling point.  

Other media employees, involved more in advertising monetisation rather than direct editorial output, have also mentioned how adopting these keyword-tweaking strategies, informed by listening to more global internet trends provided by platforms, tend to be fruitful. Small tweaks, multiplied by thousands of webpages, thanks to benefits of scale, result in interesting profits.  

However, these strategies also tend to benefit platforms’ meta-deterministic needs, with consequences on how social issues covered by journalists are phrased online. With certain keywords being chosen over others, in order to fulfil the internet’s echo-chamber of what gets attention or not, meaning, what is observed statistically as echoing what was said in the past or by others, platforms mechanically end up having quite a say as to how social issues might be framed. 

In the era of AI, this editing power based on historical patterns might be supercharged, as suggestions not only contain keywords but sometimes entire sentences. Mechanically, this also runs the risk of framing issues in a more stereotypical way.  

Re-seizing the power of metrics 

Metrics are rarely neutral tools. They benefit some and devalue others, according to decisions that are black boxed behind their deceptive simplicity. They might contain a kernel of truth, but they often reflect the interests of those who manufactured them. In the context of news, they might be one reflection of the public’s clicks, but this reflection is also heavily mediated by the infrastructure of those who are able to have an overview of the internet’s secret flows and complex behind-the-door negotiations.  

Theoretically, metrics should be seen as indicators, not as objectives. In practice however, they tend to re shape media landscapes in a way that is detrimental to media diversity, journalistic autonomy, and may contribute to journalists’ loss of meaning.  

Following a map provided by non-journalistic players may lead journalism into platforms’ projected territory, one that is biased and mostly driven by scale, where truth is defined according to statistical parameters and their own interests. That’s the price to pay when meaningful, values-oriented work is translated into simplified quantified measurements.  

But I would argue that the potential of metrics to really benefit news remains unexplored. Metrics, when seized with the awareness of what they can and cannot do, can be powerful tools to visualise injustice and quantify the reality of invisible social issues.  

When the self-fulfilling prophecy aspect of measurement is mitigated, and when metrics are seen as simple indicators, they can be used to jolt actions that favour values rather than value. Knowing the limits of metrics, and being aware of what territory is being mapped when choosing them, is therefore key to secure journalism’s interests and social mission.  

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Skłodowska-Curie grant agreement No 765140 – based at Samsa and with the University of Toulouse.

> Le Monde does 4 million video views a day against 6 million site visits. Followers are only 25% of that traffic, so YouTube works as a discovery engine rather than a loyalty channel

> Views grew 25% in 2025, driven by long formats. The deliberate choice was to invest in the newsroom and in staff journalists on camera, not a standalone TV channel or external creators

> YouTube reports 15 billion news views a year, and its revenue share leaves publishers with more than half of what they monetise on the platform

> YouTube is building paywall integration with publishers' own subscription systems, with Le Monde as a pilot partner. Today there is no way to gate video on the platform at all

> YouTube's advice: be personality-driven and direct to camera, don't copy-paste a TV format, and hire someone with real channel-management experience to train the in-house team

> SFGATE didn't add a paywall to its free site. It built a separate paid app instead, so the free web experience that built its audience stays untouched

> The name "Friends of SFGATE" beat "subscription" and "membership" in testing specifically because it read as warmer than a transaction but less formal than a cause people feel obligated to fund

> Nearly 90% of readers who took the 7-day free trial converted to paying Friends, an exceptionally high trial-to-paid rate relative to widely accepted industry benchmarks, though it's still early days

> The North Star metric shifted from total visits to repeat visits: can a reader who showed up once be brought back a second and third time, not just once by anyone

> The launch required real org change, not just a new app: a first-ever Community Engagement Program Manager role, a second video hire, and an existing editor with app experience redirected to lead app programming

> The Economist has launched a dedicated subscription tier for access to audio and video formats, priced lower than the core digital subscription to attract new audiences

> The "Halo Effect": From early signals, the team has observed an uptick in core Premium sign-ups. Seeing a cheaper, format-specific option next to the full product acts as an anchor, nudging undecided visitors to pay a little extra for the full suite.

> Net-New Audience: Conversions through Play have been driven almost entirely by brand-new readers rather than existing subscribers trading down.

> Play is part of The Economist's wider strategic development (alongside podcasts, short-form video and "Insider") towards building deeper, more "human" relationships with audiences

> According to OpinionWay, the average French person consumes about 6 media families a day, 20% already use a generative AI to get informed, but a majority say they feel overwhelmed, even fatigued, by the news

> Faced with information fatigue, unlimited access is not the answer. Les Echos is instead making the opposite bet, that of selection

> "La Sélection" consists of between 9 and 15 articles a day, chosen from a daily output of 120 to 150 pieces, following a reader-needs grid (inform, reflect, get inspired, develop, unwind) and a strict quota discipline

> Three platforms (a dedicated tab in the app, a section on the site, a daily newsletter), a weekly long read on Saturdays, and 5 shareable articles a month. Early bird price sits at €9.90 a month

> Le Monde has wanted to launch an English edition for 50+ years, but technological advances (AI translations) and digital subscriptions (delivering value internationally with ease) finally made the project possible in 2022. It's also an important move for the brand to continue to grow beyond the limited French market

> Content: Le Monde in English doesn't create original content; it exclusively translates a curated selection of the French edition (30% of all articles from Le Monde)

> Because an international reader is highly likely to already hold a primary subscription to a domestic title like The New York Times or The Guardian, Le Monde in English positions itself as a "second read," with a highly accessible pricing structure (starting at €2.50/month for the first year)

> Performance so far: around 16,000 active subscribers, 50% of whom are from the US. Approx. 5 million monthly visits, and subscriptions are the core business driver, representing 66.2% of revenue, followed by licensing (16.9%), B2B (12.3%) and advertising (4.6%)

> "It's not about translation": success comes down to content market fit, product distribution, and understanding the distinct profile of your international audience. And whilst AI has driven productivity gains on this front, it's also created journalism jobs - AI handles the mechanical translation, freeing human journalists to focus purely on editing, nuance, and curation

> The Sun launched Sun Club in 2025 as a £1.99 freemium membership, layering subscription revenue and first-party reader data on top of an already large free audience

> A year in, 65% of members are women, and TikTok has opened a genuinely new audience: those subscribers are 85% female and 15 years younger than the average member

> First-week app activity is the strongest predictor of long-term retention. The most loyal members read over 100 articles a week, almost entirely in the app

Lessons:
> Value-driven perks are the core of the proposition, and for this audience they're the retention tool that matters

> Freemium can serve advertising and subscription goals at once, because a known, logged-in audience strengthens both