Brian Morrissey & Daniel Kolitz

May 13, 2025 — 9 min read

Summary

Audience data is the foundation of modern media. The future of the media business lies in understanding their audiences and better serving them. This is fundamental for driving growth.

But for many publishers, it’s still a pile of parts, scattered across systems, inconsistently owned, and underutilized. In an era of tighter budgets, disappearing third-party signals, and rising AI capabilities, publishers are rethinking how they collect, structure, and activate audience data. And they’re increasingly turning to AI to help do it.

This report, based on a survey of 79 media professionals and interviews with product and data leaders, explores how AI is being used not just to generate content or optimize workflows, but to strengthen the infrastructure behind direct audience relationships. The findings point to a shift: AI is less about automation for its own sake and more about enabling smarter audience strategy through improved segmentation, personalization, testing, and predictive modeling.

However, progress remains uneven. This shift is taking place in a context where publishers are under financial pressure that has eroded investment levels. Most publishers are still in early-stage experimentation. Teams often lack the skills and systems to operationalize AI at scale. And while efficiency gains are real, few organizations have connected those wins to deeper personalization or improved lifetime value.

“There’s a million companies out there that can do amazing things for you,” Time COO Mark Howard told us. “You need to figure out, with limited resources, how to make it work for you.”

Still, the direction is clear. AI is becoming embedded in audience workflows. Vendor expectations are shifting. And the most effective publishers aren’t chasing bold bets—they’re solving foundational problems with better tools and clearer ownership.

Publishers know that audience data is the foundation for long-term growth, but most still struggle to turn that belief into an operational reality. Data often sits across multiple systems with little coordination, and no single team owns the end-to-end strategy. AI tools offer a potential path forward—but without strong ownership and aligned processes, they only add another layer of complexity.

• 33% of publishers cite siloed data across platforms as one of their biggest challenges.
• 22% say no single team owns their audience data strategy.
• 35% cite difficulty deploying AI-derived insights at scale.
• 33% point to a lack of technical resources or skills.

Publisher POV:

• “There’s a lack of shared knowledge, awareness, and access to the data. Teams often don’t even know what’s available or how to use it. That disconnect makes it almost impossible to activate what you have. You end up with analysis sitting in one part of the business and activation happening somewhere else.” — Head of product at a vertical publisher
• “Even asking the right question requires fluency—and most people aren’t fluent. If you’re not close to the data, it’s really hard to even formulate what you’re trying to find. And when people do ask, they often lack the context to interpret the answers. This leads to surface-level insights instead of strategic use.” — Gilad Lotan, head of AI, data science and analytics, BuzzFeed
• “The problem has been that to take advantage of AI in a scaled way with your staff, you need a pretty nuanced AI policy. It can't just be ‘it's forbidden’ or ‘you can only use this one service.’ Once you open the door, there's going to be some error. And even with technical safeguards, the policy layer matters just as much as the tools.” — Pete Pachal, AI analyst

BlueConic POV:

“The problem isn’t that publishers lack data—it’s that no one owns the strategy. Teams are sitting on mountains of behavioral and identity data, but without coordination, it just adds complexity. What we see working is when companies centralize ownership and start applying AI not just to analysis, but to actual activation.” — Will Barker, principal customer success manager, BlueConic

💡Key Takeaway:

Publishers recognize the value of audience data, but they still lack the alignment to make it actionable. Without ownership and interoperability, AI can’t unlock the potential sitting in siloed systems.

Making sense of the data

As an organization, BuzzFeed is drowning in data: every action a user takes on their website is logged and indexed. This data powers everything they do—their algorithms, their internal and external reports, even their content.

Given the sheer quantity of signals coming in, scaling this operation was difficult at first. The company eventually managed it, supplanting statistical sampling methods with systems that ensure a single source of truth—but another problem remained: While the company's analysts were able to get a lot out of this data, everyday employees were struggling to make sense of it.

"Many people know how to read a spreadsheet, or pull a number from a specific dashboard," said Gilad Lotan, Head of AI, Data Science and Analytics at BuzzFeed. "But the interpretation part—that's not easy."

Lotan uses the example of a product manager. If a product manager wants to put together a strategy for the upcoming quarter, they can simply pull numbers from an existing dashboard. But anything more complex—for instance, understanding user behavior by segmenting users into cohorts—would typically require the help of an analyst.

AI, it turns out, has significant uses here. Large language models (LLMs) allow users to look not only for keywords but for text with similar meanings—as Lotan puts it, "the actual words don't have to appear in the text for you to find adjacent text." This helps not only with data interpretation among non-technical personnel, but with a range of tasks—including BuzzFeed's internal recommender systems. "It might not be exactly what you were hoping for," said Lotan, "but it’s better than a blank page, right?"

AI is beginning to reshape how publishers handle fragmented tech stacks. Instead of layering more tools on top of outdated infrastructure, teams are using AI to help unify and translate their existing systems—surfacing insights, ranking audience segments, and triggering actions that were previously out of reach. For many, AI is less about automation and more about orchestration.

• 58% of publishers are using or planning to use AI for audience insights.
• 48% are applying AI to segmentation or churn reduction.
• 39% say audience insight is a top vendor selection factor.
• 35% say operationalizing AI insights is still a challenge.

Publisher POV:

• “The performance is better. When you stop paywalling users unlikely to convert, everybody wins—readers, marketers, and journalists. It changes the value equation when you can identify and act on patterns in real time.” — Product lead at B2B publisher
• “We have systems that classify and structure information at scale. That makes it possible to detect patterns we wouldn’t otherwise see. You can train these systems to surface relationships in the data that aren’t obvious, especially when you’re dealing with massive volume. It changes the scale and speed of insight.” — Gilad Lotan, BuzzFeed

BlueConic POV:

“What we see is that publishers often have the data they need to build better audience segments—but they’re stuck manually translating it across systems. AI becomes most powerful when it’s connected to activation. That’s the unlock: not just generating insight, but operationalizing it quickly.” — Will Barker, BlueConic

💡Key Takeaway:

AI is helping publishers make their stacks more coherent. It’s allowing teams to operationalize insights across silos—but it only works when infrastructure is aligned.

AI is not yet a source of strategy for most publishers—but it is a source of speed. The clearest short-term gains come from eliminating repetitive tasks, whether it's optimizing headlines or summarizing interviews. These wins matter, especially for lean teams, but they rarely translate into broader audience impact.

• 59% say AI has improved efficiency.
• 53% use AI for A/B testing.
• 45% use AI for SEO optimization.
• Only 13% say those gains have been “significant.”

Publisher POV:

• “AI for social copy, SEO, headlines—that’s table stakes now. For a lot of team members, it's the most persuasive use case: ‘Hey, this thing you never wanted to do in the first place? An email bot can do most of it.’ It’s not revolutionary, but it’s extremely practical, and that’s what gets people to actually use it.” — Pete Pachal, AI analyst
• “We’ve always been looking for efficiencies. That was the early use of AI, and it’s worked out. I mean, I think transcriptions are still incredible with AI. The key is, we always like to flag occasionally when something isn’t exactly what was said, and we remind some of our newer reporters not to assume everything is going to be perfect.” – B2B publishing executive

BlueConic POV:

“Where we’ve seen the biggest gains so far is in fine-tuning paywall logic and helping publishers decide what kind of offers to show which users. That’s not glamorous AI—but it’s measurable. And it directly improves conversion without adding much workload to the team.” — Will Barker, BlueConic

💡Key Takeaway:

Efficiency is where AI is gaining traction first. These wins are necessary, but not sufficient—they must lead to bigger strategic growth outcomes.

Case Study: Striking AI deals

Decades into the digital age, few outlets can match the name recognition of TIME Magazine. Accordingly, their approach to GenAI was going to be closely watched by industry observers. According to their COO, Mark Howard, they had three options.

They could do nothing—but of course that wasn't an option for them. Since 2018, TIME has been owned by Salesforce founder Marc Benioff—one of the key figures driving the AI revolution. This is someone who believes that AI agents are the future not just of business generally but of journalism specifically—ignoring this technology wasn't on the table.

They could, like The New York Times, litigate—but there was little appetite interally for that.

Which left one remaining option: negotiate.

For a company of TIME's size—a few hundred employees, down from thousands at the peak of the print era—this has seemed like the wisest approach. From a financial perspective, it's not transformational—but, per Howard, it isn't about the money.

Companies are understandably fearful of competition from ChatGPT (why seek out and skim an article for information when you can just ask a bot exactly what you want to know?). The wisest approach, then, would be to secure your content pride of place within those ecosystems. "It puts us in the inner-circle with those companies," said Howard. "It gets us executive access. It gets us access in most cases to quarterly VRs and roadmap presentations. We get to have a seat on the inside, versus being on the outside trying to figure things out."

Personalization is one of the most frequently cited goals of AI adoption in media—but it remains one of the most elusive. Most publishers are still stuck at cohort-level segmentation. Moving to true 1:1 personalization requires not only the right tools, but trust in their output and alignment across platforms.

• 53% are using or plan to use AI for personalization.
• Only 5% say they’re “very effective” at it.
• 42% rate their personalization performance as poor.
• 35% say they can’t operationalize AI outputs.

Publisher POV:

• “We’re still working with cohorts. We’re not close to truly dynamic customization. Right now, most of what we consider personalization is really cohort-based targeting—if you're in a certain group, you get a slightly different experience. Our goal is true one-to-one personalization, but that requires a level of data maturity and infrastructure we’re still building toward.” — Gilad Lotan, BuzzFeed
• “AI hasn’t caught up to the hype. Everyone thought it would just spit out perfect content with minimal prompting, but there’s almost always a need for a human in the loop. That’s especially true with personalization, where voice, tone, and nuance are everything. The more templated the use case, the more likely it is to succeed—but that’s not where most publishers are focused.” — Pete Pachal, AI analyst

BlueConic POV:

“You can’t personalize in real time if you’re relying on batch data. We’ve seen teams unlock real impact by moving from email-based personalization to on-site recommendations that adjust dynamically based on behavior. That shift requires trust in the system and tight integration across platforms.” — Will Barker, BlueConic

💡Key Takeaway:

Efficiency is where AI is gaining traction first. These wins are necessary, but not sufficient—they must lead to bigger strategic outcomes.

AI's technical capabilities are advancing quickly—but organizational readiness is lagging. Mistrust in AI-generated results, limited internal expertise, and a lack of formal governance are keeping publishers in a perpetual pilot phase. Culture, not tooling, is the biggest blocker.

• 61% cite lack of internal expertise.
• 46% say they mistrust AI outputs.
• Only 13% feel “very confident” in their team’s AI readiness.
• 39% cite unclear ROI as a leading concern.

Publisher POV:

• “There’s a million companies out there that can do amazing things for you. And you need to figure out, with limited resources, how to make it work for you. There’s both the high-level analysis—how do we weed through everything that’s out there—and real opportunity.” — Mark Howard
• “Everyone’s finding their own comfort level. That’s part of the process, but it also means that there’s no consistency. One team might be using AI to summarize audio while another avoids it altogether.” — Head of product at a vertical publisher
• “The goal is to have this kind of tooling integrated directly into your workflow, like SEO suggestions in Slack or the CMS. Not everyone’s there yet, but there’s a definite push in that direction and broad acceptance.” — Pete Pachal, AI analyst

BlueConic POV:

“There’s often a sense of, ‘I know what my audience cares about, and I want to decide what shows up on the homepage.’ But when you’re publishing thousands of articles, it’s not scalable for an editor to manually choose what shows up at the end of every article. There's concern that automated recommendations might surface irrelevant or awkward content.” — Will Barker, BlueConic

💡Key Takeaway:

AI won’t scale until organizations build the trust, training, and governance needed to use it with confidence.

AI is a must-have in vendor conversations—but it's far from a must-fund. Most publishers are still experimenting with low-stakes integrations and waiting for clearer ROI before ramping up budgets. In this phase, restraint is a feature, not a flaw.

Key Data Points:

• 78% say AI is important in vendor selection.
• 65% plan to increase investment in the next year.
• 50% spend less than $10K annually on AI.
• Only 5% spend more than $200K.

Publisher POV:

• “When we talk to any vendor, I always say, ‘OK, for your product to work the way that you're selling it, what would the ideal structure of our team be?’ And some of them are like, ‘You should probably have three people managing this.’ And it’s like, ‘OK, talk to you later.’” – Head of product at a vertical publisher
• “What I’ve learned is that everyone’s looking at it now. So if you think you’ve found a trick to leverage, I’m sure everyone else is doing the same. Don’t rely on tricks. This is what happened with Facebook traffic.” – B2B publishing executive

BlueConic POV:

“Publishers are now asking: how can I identify the right kind of offer for different people? Maybe one person needs a three-month trial at 99 cents, someone else needs six months... That’s where modeling becomes key.” — Will Barker, BlueConic

💡Key Takeaway:

Investment is coming—but deliberately. Most publishers are using AI to prove small wins before making larger bets.

Conclusion

AI is no longer a curiosity or a buzzword for publishers—it’s becoming a baseline capability. But even as tools improve and vendor offerings expand, the real work is happening under the hood. The biggest gains aren’t coming from futuristic deployments, but from incremental fixes to broken data workflows and underpowered audience strategies.

Across the industry, publishers are moving toward a new operating model: one that treats AI not as a standalone solution, but as an accelerant to foundational change. From streamlining testing to supporting segmentation, AI is showing up first in places where the pain is greatest and the lift is lightest.

But deeper transformation remains elusive. The full potential of AI—dynamic personalization, predictive engagement, lifecycle optimization—depends on systems, governance, and organizational buy-in that most publishers are still building. Siloed teams, inconsistent ownership, and low internal trust are keeping many efforts stuck at the surface.

Still, momentum is real. Publishers are investing. Teams are experimenting. And more importantly, they’re aligning AI investments with real business needs rather than novelty. The most successful cases aren’t chasing headlines—they’re solving for clarity, consistency, and speed.

The road ahead isn’t about replacing people with platforms. It’s about enabling teams to move faster, make smarter decisions, and act on data that has too often gone unused. AI can help publishers get there—but only if they fix the foundation first.