Beyond Traffic:
personalization, trust and new ways of connecting with audiences
Opening
From reach to connection
For years, media organizations measured their relationship with audiences in terms of reach: how many people arrived at a story, how long they stayed, what percentage clicked. Those metrics organized newsrooms, oriented editorial strategies and defined criteria for success. Artificial intelligence arrived in an ecosystem where that logic was already exhausted. The audiences media seek today are more fragmented, more demanding, harder to retain and, at the same time, more willing to engage when they find something they genuinely consider relevant.
The change is structural. For more than a decade, digital platforms trained people to consume personalized, immediate, on-demand content. Spotify, Netflix, TikTok and recommendation algorithms built a new standard of experience that journalistic media did not ask for, but that their audiences internalized. Today, someone who reads a newspaper or listens to a news podcast carries with them expectations formed by systems that learn from every interaction. Competing in that environment requires understanding that the benchmark is defined by any digital experience that competes for the same time and the same attention.
What further complicates this scenario is that fragmentation has also reached trust. Data from the Reuters Institute Digital News Report shows that active interest in news has been declining steadily since 2017 in most surveyed markets, with a phenomenon researchers named "news avoidance" — people stepping away from information out of exhaustion, anxiety or a sense of powerlessness in the face of what they consume. In that context, producing more content deepens the problem.
Faced with that double pressure, artificial intelligence enters the conversation about audiences from two opposing angles. On one hand, as a tool to scale personalization, accelerate distribution and optimize content relevance for each segment. On the other, as a risk factor for trust, insofar as its accelerated and opaque adoption feeds the perception that media organizations prioritize efficiency over editorial judgment and genuine connection with their readers. Navigating that tension with clarity is the central challenge this chapter sets out to explore.
Distribution
The collapse of the distribution model
Since media organizations began their digitalization process, now thirty years ago, traffic from search engines first and social networks later was the engine that organized much of their distribution strategies. Publishing more, optimizing for Google, distributing on Facebook and Twitter and measuring reach in page views became the dominant model — predictable and scalable. Newsrooms learned to work with that logic, and many built their business structures around it.
That model entered collapse between 2022 and 2025, hit by a perfect storm of mutually reinforcing changes. Meta drastically reduced the organic distribution of journalistic content on its platforms while Google began answering questions directly in its search results — first with featured snippets and then with AI-generated answers — reducing the need to click through to a story. TikTok and YouTube recommendation algorithms accelerated the shift toward entertainment content at the expense of news, and the emergence of conversational assistants introduced a new way of accessing information that bypasses media as intermediaries.
The impact was immediate and uneven. Media most dependent on referral traffic saw their visits decline steadily. Some responded with more volume, betting that scale would compensate for the per-piece drop. Others began to reconsider from where they were building their relationship with audiences. The difference between both responses is strategic. The former kept betting on reach; the latter began betting on connection.
What the collapse of the distribution model made clear is that traffic was never synonymous with audience. A visit is not a reader, a click is not a commitment. A user who arrives from a search and abandons an article after thirty seconds builds no relationship with the outlet that received them. For years, that distinction was available in the data but was ignored by the logic of advertising monetization, which rewarded volume above depth.
The response that gained ground in newsrooms of different sizes and regions points in a different direction. Newsletters, apps, communities, memberships and restricted-access products are the most concrete expressions of that reorientation. In every case, the common denominator is the same: the relationship with the audience depends on an active decision by that audience, not on a third-party algorithm. Artificial intelligence participates in this new model in ways still being defined — from content personalization to automation of direct distribution flows — but its value depends on there first being an audience willing to sustain that connection. And that willingness is built with editorial judgment, consistency and trust.
Personalization
Personalization without losing editorial identity
Personalization is one of digital journalism's oldest promises and also one of its worst executed. Since the first experiments with recommendation algorithms, the industry held that offering each reader the most relevant content for their interests would increase engagement, reduce churn and strengthen the bond with the outlet. In practice, what was implemented in most cases was an impoverished version of that idea — limited to showing more of the same, reinforcing topics that already generated clicks and avoiding the friction of exposing audiences to content they had not actively sought.
Artificial intelligence significantly expands the technical capabilities for personalization. Current systems can analyze behavioral patterns, infer preferences, adapt formats, adjust sending frequencies and generate differentiated versions of the same content for different audience segments. That capability, well directed, makes journalism more accessible and relevant for those who have already chosen to consume it. But for that direction to be possible, a starting point is needed that technology alone cannot provide — understanding why each person consumes news.
The user needs model, developed in the public service journalism sphere, offers a useful framework for thinking about that question. People approach news for very different reasons: to stay informed, learn something new, solve a concrete problem, find community, validate a perspective or follow a developing story. Understanding those motivations changes how content is designed. It is no longer about producing a story that is clear and current, but about answering a specific reader need at a given moment and on a given platform.
The problem arises when personalization becomes the organizing criterion for the entire informational experience. An outlet that only shows each user what they already find interesting abandons one of its most historically valuable functions — putting in circulation information that readers need even if they have not sought it. A society's informational agenda cannot be built exclusively from individual preferences. There are stories that matter even if they generate fewer clicks, topics that require visibility even if they produce more discomfort than engagement, and perspectives that broaden the framework of understanding even if they do not correspond to each user's habitual consumption profile. Delegating those decisions to an algorithm is an editorial renunciation that no personalization system can compensate for.
This is the point where each outlet's identity becomes a strategic asset. Journalistic organizations that are clear about what kind of journalism they do, for what community they do it and with what editorial criteria they produce it are better positioned to use personalization as a tool without being redefined by it. Personalization should operate within the limits that editorial identity establishes. When and how to show content can be an algorithmic decision; what content to produce and why remains a human decision.
The newsrooms advancing most solidly in this area are those that incorporated journalistic values as explicit criteria within their personalization systems, so that technology reinforces the editorial commitment rather than eroding it. Ethical personalization is compatible with efficiency when systems are designed with those values incorporated from the start — not as part of a later correction. The decision to use them in that direction, oriented toward genuine connection with audiences rather than maximizing screen time or reducing subscription cancellation rates, is an editorial decision. And like every editorial decision, it reveals what kind of outlet each organization wants to be.
New Formats
Conversational journalism
One of the most significant shifts introduced by artificial intelligence is still incipient, but its trajectory is becoming clear. Audiences are moving from searching for information to conversing with information. That displacement — from link to summary and from click to direct response — reconfigures the place media occupy in their readers' informational journey. Conversational assistants operate as a new interface for accessing information. In that ecosystem, the relevant question for media shifts from how to appear in the top search results to how to be present at the moment someone formulates a question and expects something — or someone — to answer it.
Some media organizations have begun exploring the development of their own conversational formats. Editorial chatbots that allow browsing archives, assistants that answer questions about ongoing coverage, newsletters structured as dialogues, question-and-answer formats that organize information from the reader's perspective. These experiences, still experimental in most cases, share the same intuition: if audiences have grown accustomed to interacting with information conversationally, journalism can find there a way to build presence, relevance and direct connection — even though that requires activating resources many newsrooms possess without quite knowing how to use.
In that context, the journalistic archive emerges as one of media's most valuable and least exploited assets. Semantic search tools and language models are changing the equation. What for decades was an inaccessible repository can become conversational infrastructure — a knowledge base that readers can explore, consult and activate.
The strategic value of this material lies in something no external model can replicate: local coverage, distinctive historical perspectives and an editorial voice accumulated over years. In an environment where conversational assistants compete with media to answer questions about current events — consuming journalism without attribution — the own archive is one of the few territories where media have a genuine structural advantage, even though translating that value into product remains an outstanding debt for most newsrooms.
The opportunity exists, though the risk of repeating the same mistakes deserves attention. Journalism has spent years chasing platform changes — first SEO, then social networks, then vertical video. In each case, adapting to external distribution logic ended up eroding something of the original editorial proposition. With conversational environments, the temptation is similar: optimizing content to be cited by language models, structuring information to facilitate automatic extraction and writing for machines before people, with the risk of losing once again the direct connection with audiences that took so long to build.
What distinguishes a solid conversational strategy from a superficial adaptation is, once again, editorial judgment. Media that build their own conversational formats — anchored in their values, their voice and their relationship with a specific community — are betting on something no language model can replicate: a recognizable perspective, accumulated trust and sustained presence in their readers' informational life. That bet connects directly to the dimension that conversational journalism still needs to resolve: how to sustain and communicate trust in an ecosystem where more and more content arrives without the reader quite knowing where it came from.
Transparency
Trust, transparency and labeling
Trust between a media organization and its audience is built over years and can erode within weeks. AI added a new dimension to that risk — one few journalistic organizations anticipated. By incorporating generative systems into their production workflows without communicating it clearly and without designing use policies, many media organizations fed the perception that something had changed in how information was produced, even if no one said so openly. That perception, however diffuse, has concrete consequences for audiences' willingness to trust what they read.
The problem deepens because that silent adoption coincided with a pre-existing credibility crisis in media. Audiences were already arriving with elevated levels of distrust, and the proliferation of synthetic content in the broader information ecosystem added a new layer of uncertainty. In that context, the question many people began asking themselves about any piece of content — whether produced by a person or generated by a machine — stopped being a technological curiosity and became an everyday criterion for evaluating journalism's credibility.
Transparency about AI use operates in that landscape as an active editorial tool. Media organizations that communicate clearly when and how they use artificial intelligence in their production processes build a trust differential in an industry where that clarity is still scarce. That communication can take different forms depending on the type of content and the context of use: editorial notes explaining the production process, labels identifying the degree of algorithmic intervention in a piece, or public policies that detail usage criteria at an institutional level.
The labeling of AI-assisted or AI-generated content concentrates one of the most active debates and, at the same time, one of the least resolved. The questions running through it are more complex than they appear at first glance. What threshold of algorithmic intervention justifies a label? Is an automatic transcription labeled the same as an article written entirely by a generative model? Does labeling inform the audience or confuse them? Answers vary by outlet, regulatory context and the specific relationship each organization has with its readers. What does seem clear is that the absence of labeling in environments where audiences suspect it but do not know for certain erodes more trust than any imperfect transparency format.
That absence also has a collective dimension that transcends individual decisions by each outlet. The proliferation of synthetic content produced by actors with no commitment to accuracy or attribution installs in audiences a structural distrust that journalistic media end up absorbing even when their own use of AI is responsible. The contamination of the information ecosystem is one of the strongest arguments for media with declared editorial commitments to invest in distinguishing themselves actively — communicating their usage criteria with the same energy with which they communicate their coverage.
Trust is no longer an implicit value assumed through institutional track record. In the era of generative AI, it is also built with explicit, sustained and verifiable transparency about how information is produced. And that construction requires editorial decisions as deliberate as those that define any other dimension of journalism.
Action
From reading to action: nine orientations for transforming the relationship with audiences
Thinking about the relationship with audiences from a different place is the first step. The second is knowing where to start. The nine orientations that close this chapter bring together reference frameworks, questions to circulate among teams and concrete actions designed so any newsroom can move forward with available resources and from its own priorities.
01
Your community's questions
A concrete editorial exercise for building relevance in an ecosystem where AI platforms have become new gateways to information is to identify thirty real questions your audience is asking and answer them with clarity, precision and utility. The methodology starts by identifying real demand through reader service queries, internal search data, social media comments and trend analysis. Those answers can become articles, newsletters, WhatsApp threads or short service podcasts — and over time, a knowledge base that trains the outlet's own assistants.
Guiding questions
Which topics does your audience have recurring questions about that your outlet could answer better than anyone? Is there a channel where those questions arrive directly — reader service, social networks or comments? What formats would be most useful for distributing those answers beyond the traditional article?
Choose one section of your outlet and list the ten most frequent questions your audience asks about that topic. Answer the first one this week.
02
Audience routines
Informational habits are built in everyday interstices — the morning coffee, the commute, the midday break. Media organizations that manage to be present in those moments with something useful and clear build a relationship that no external distribution algorithm can replace. The focus shifts from seeking a single dominant format to creating multiple touchpoints adapted to different contexts and levels of attention. The INMA product blog goes deeper into this approach with concrete examples of how different media organizations are thinking about habit as a strategic axis.
Guiding questions
At what times of day does your audience seek information and with what level of available attention? Which products or formats does your outlet have that fit into those routines, and which create friction? What devices does your audience use at each moment and what does that mean for content format and length?
Choose a segment of your audience and map their informational journey during a typical day. Identify three moments when your outlet could be present but currently is not.
03
Change as the goal
Change-centered journalism proposes a shift from a model organized around content and volume to one that places audiences and the changes information can drive in their lives at the center. Applied to the relationship with audiences, it means designing coverage with a clear recipient, a concrete problem to solve and a possible action at the end of the informational journey. The framework was developed by Jazmín Acuña, co-founder of El Surti in Paraguay, in a paper published by the Reuters Institute.
Guiding questions
What concrete change do you want to generate in your audience's life with the coverage you are producing? How would you know if that change occurred? What kind of journalism does your outlet produce that generates a real and verifiable impact on its community?
Take a recent piece of coverage and rewrite its objective in terms of audience impact. What should someone who read it be able to do or know?
04
From audience to community
The difference between a subscription and a membership seems simple but profoundly reorganizes how you think about the relationship with your readers. A subscription is transactional — you pay and receive a product. A membership means joining a cause because one believes in the importance of the work being done. That difference organizes editorial, design and communication decisions. The thick membership model is the one that builds more lasting bonds — members participate in events, offer feedback, respond to calls and contribute their knowledge to the journalistic process. Media organizations growing most in this model are those that found concrete ways to integrate that participation into their daily work. The connection with AI arises naturally here: if an active community contributes perspectives, data or experiences, AI can help process that input at scale so journalists can work with it without drowning in volume. This approach was systematized by Jay Rosen and his team in the Membership Puzzle Project, a four-year study in newsrooms on five continents.
Guiding questions
What spaces does your outlet have for the audience to contribute something beyond their attention? Is there any research or coverage process where your community's knowledge could improve the final result? How easy is it today for someone in your audience to contribute something concrete to your journalism?
Identify a recent piece of coverage where you needed information your audience probably had. Design a simple call-out for the next similar story and measure how many useful responses you receive.
05
User needs
Identifying why each person consumes news changes how any piece of content is designed, headlined and distributed. People do not approach information for a single reason or with the same level of attention at all times of day. Some want to stay informed, others want to learn something new, solve a practical problem, find inspiration, be entertained, find community, validate a perspective or follow a developing story. Understanding which of those motivations underlies each piece changes basic editorial decisions — from the headline to the distribution channel. The framework was developed by Dmitry Shishkin during his time at BBC World Service, and the full documentation of the user needs model is available on the Smartocto platform.
Guiding questions
What concrete need does each of the informational products your outlet produces set out to meet? How does that need change depending on the channel through which the content arrives? Are there needs of your audience that your outlet identifies but does not yet know how to respond to?
Apply the model to three recent pieces from your outlet. Identify what need each one resolves and whether the chosen format is the most appropriate for that need.
06
Public value as editorial compass
Newsrooms have spent years justifying their value with individual metrics — clicks, dwell time, engagement. Those figures are useful in certain contexts, but they rarely answer what matters most: what place journalism occupies in people's lives and in a society's health. The Public Method, developed by Dutch public broadcaster NPO and the Sjoege collective, proposes a different approach. Its core is a rose-shaped model that crosses the individual needs of audiences with the functions society assigns to journalism — uncovering information, setting the agenda, representing diverse voices and holding power to account. Change appears on the outer edge not as an objective but as a possible consequence of purposeful journalistic practice. NPO already uses it to define editorial products and portfolio strategies, combining quantitative data with qualitative reflection and editorial judgment.
Guiding questions
What social function does each informational product of your outlet serve, beyond its performance in metrics? Are there voices in your community that your habitual coverage does not represent? What societal change can be attributed to the journalism your newsroom produced in the past year?
Take your outlet's product portfolio and place each one in the rose model. Identify which social functions are covered and which are absent. That map is the starting point for a conversation about editorial purpose that no metric can replace.
07
Less volume, more relevance
While AI enables producing more content with fewer resources, the underlying problem points in the opposite direction. As developed in the first section of this chapter, the news avoidance documented by the Reuters Institute shows that the problem is not a lack of information but its excess. The same phenomenon has its internal correlate in newsrooms — a tension between maximizing attention and respecting audience time that many editors recognize but few organizations resolve explicitly. Emily Goligoski's work at The Atlantic points to the same problem from another angle: audiences who feel their experiences are taken seriously are more likely to remain engaged and pay. Publishing less but with greater relevance is the most appropriate response to an ecosystem where attention is the scarcest resource and where AI is already generating more noise than any newsroom can manage.
Guiding questions
How many of the pieces your outlet published last week responded to a real audience need, and how many responded to the inertia of the news cycle? What would happen if you published half as much and dedicated the remaining time to deepening what you did publish? What type of content from your outlet has the most value for your audience and the least presence in your weekly planning?
Take a recent week's output and classify each piece according to whether it resolved something concrete for your audience or simply covered a topic. Use that map as input for the next planning meeting.
08
Connection metrics
A newsroom that changes its audience strategy but keeps measuring with the same indicators will quickly return to making the same decisions. The first step is agreeing internally on what signals indicate that the relationship with the audience is strengthening. Return rate, reading depth, sustained newsletter open rates, community participation and the conversion of occasional readers into regular subscribers speak to a relationship sustained over time.
Guiding questions
What indicators is your newsroom watching today and what editorial decisions does that focus generate? What signal of real connection with your audience are you still not measuring? What does your team celebrate as success, and what does that tell us about the journalism we are prioritizing?
Identify two recurrence or depth indicators your outlet could start tracking this week and present them at the next team meeting.
09
The archive as editorial heritage
A media outlet's archive is often its most underexploited asset. Before designing any conversational experience or audience product, it is worth doing a simple mapping of what own coverage exists on the topics that matter most to your community, which series or historical investigations could have a second useful life and what data or own records exist in no other source. That inventory usually reveals an editorial heritage that the daily urgency keeps invisible but that represents a genuine structural advantage over any content producer arriving later.
Guiding questions
What does your archive hold that no other outlet in your region could offer? What historical coverage might be useful or relevant to your audience today? What internal barriers prevent your archive from becoming an active resource for the newsroom and audiences?
Choose a central topic in your regular coverage and review what own material exists on that topic over the past five years. That inventory is the starting point of any archive activation strategy.