How Publishers Can Build a Lasting Data Advantage in the Age of AI
As AI reshapes media buying, Nick Henthorn of InfoSum explains how publishers can use first-party data more effectively without giving up control.
Publishers have spent years building detailed first-party data around audience interests and content consumption. The next challenge is connecting those signals with advertiser, retailer and broadcaster data without moving raw customer information into shared systems.
That matters as AI changes media planning and buying. Advertisers increasingly want richer audience intelligence to guide budget allocation, while publishers need to protect the data that makes their inventory distinctive.
In this Q&A, Nick Henthorn explains how decentralised data collaboration can help publishers create more useful advertising products without surrendering control of their audience data.
As AI becomes ubiquitous, how can publishers use their unique data assets to give themselves a competitive edge?
As AI solutions and agentic workflows are increasingly involved in media planning and buying, it’s critical for publishers that their data is visible within these systems. Advertisers and agencies want access to signals that augment their customer understanding and help them understand where best to allocate media budgets. Through data collaborations, premium publishers can realise the value of their first-party data, helping advertisers and agencies find the right insights and quality media environments for their campaigns.
By ensuring these collaborations are easy to implement and allow all parties to protect their data, publishers can take a big step forward toward a future-proof revenue stream.
There’s further value that can be realised from first-party data when connecting it with signals from complementary partners. A publisher knows a great deal about how their audiences consume content. They know which topics drive engagement and how long readers spend on a page. They see what kind of content draws someone back repeatedly. What they typically don’t know is what those same readers are buying or what they watch on television. They don’t have a view of how audiences respond to advertising outside of their own environment. This is where secure data collaboration can make further impact.
When a publisher connects its audience insights with a broadcaster’s viewing data or a retailer’s purchase signals, AI models built on the findings can identify patterns that no single party could see alone.
Publishers who do this create audience intelligence specific to their portfolio and their partnerships. Competitors working from generic data sources can’t replicate it because the underlying combination of signals is unique to each collaboration.
In a collaborative AI environment, how can publishers ensure their proprietary audience data doesn’t inadvertently benefit their competitors?
This is one of the questions we hear most from publishers, and it’s an entirely reasonable concern. The worry is that once your data enters a shared system, you lose control over how it’s used. In an AI context, that concern is amplified because data fed into a model doesn’t just inform a single query, but also shapes the model’s behaviour in ways that are hard to track and harder to reverse.
The answer lies in the architecture of the collaboration itself. Data should never move. Instead of sending your data to a central environment where an AI model trains across it, the model should come to your data. Each publisher keeps their data within their own cloud or technical infrastructure, and queries run across the network under permissions they set and control.
This means a publisher can participate in collaborations to power AI models without exposing raw audience data to any other party. Your competitor cannot access your data through a shared system, even accidentally, because your data never enters that shared system in the first place. You define which partners can run queries against your data and what types of analysis are permitted. A full audit trail is maintained throughout.
How does data collaboration help publishers solve the ‘fragmented view’ problem for their advertising partners? Do you have any specific examples?
Advertising partners often see a publisher’s audience through a very narrow window. They can see that a particular user visits the travel section, but they have no way to connect that behaviour to purchase intent or a financial profile. Each publisher represents one piece of a much larger picture. But advertisers want the whole picture, and data collaboration lets publishers address this demand directly – without surrendering the data that makes their audience valuable in the first place.
For instance, a publisher whose readers skew towards home improvement content could connect their data with that of a major DIY retailer. Using privacy-preserving analysis, both parties can identify the overlap between editorial engagement and purchase behaviour, without either party handing over their customer records.
The advertiser gets a far clearer signal about which audiences are genuinely in-market. The publisher can demonstrate that their readers are active purchasers, not passive content consumers.
Another example could involve a news publisher connecting their data with a broadcaster’s addressable TV audience to show an advertiser precisely which programmes their most engaged readers watch. That kind of cross-media insight helps advertisers plan more effectively and reduces the guesswork around where a campaign should run.
What role does InfoSum’s technology play in simplifying the technical and governance hurdles for publishers ready to scale AI?
One of the most common challenges is the operational overhead. Setting up a new data partnership with legacy centralised technology requires extensive legal reviews and privacy assessments before any technical work can begin due to the requirement to share data. By the time a publisher has cleared those hurdles with one partner, scaling to five or 10 more is genuinely difficult.
The decentralised technology we’ve built at InfoSum eliminates the need to share data and therefore removes that friction. Beacons, our AI-ready collaboration infrastructure, deploys our platform directly within a publisher’s own cloud environment. The publisher’s data always stays exactly where it resides. There’s no migration and no centralised data pool that other parties can access. For example, a publisher using Google Cloud can connect with a partner running a different infrastructure without either party moving a single record.
Once that foundation is in place, governance is far more manageable. Publishers can pre-clear specific data types that have been approved by governance teams, such as contextual signals, behavioural insights or audience segments, removing the need for new approval requests with each new partnership. That dramatically reduces the time between identifying a collaboration opportunity and actually running analysis. For publishers managing a large number of advertiser and partner relationships, this means scaling without increasing the compliance burden at the same time.
Why is ‘non-movement of data’ the essential foundation for sustainable governance?
AI governance is a relatively new discipline for many organisations, but publishers who have been thinking carefully about data privacy for years will recognize the underlying principle. The safest data is data that stays under your full control.
When data moves into a third-party AI system, that data becomes absorbed into the model, and recovering full control of how it’s used becomes very difficult. The organisation running the AI system usually also takes on responsibility for how your data is used, rather than the publisher who collected it. Both of these outcomes carry real risk.
The non-movement principle resolves this. If AI analysis runs against your data where it lives, rather than pulling your data into a centralised model, you retain full ownership throughout. You can audit what was queried and revoke a partner’s access. Your audience data stays outside any model that could quietly benefit a competitor. This is also important for regulatory compliance, as data protection laws in most markets require organisations to know where personal data sits and who can access it.
Publishers who build their AI strategy on this foundation are protecting themselves commercially and legally. The publishers who don’t will face increasingly difficult questions from their own governance teams and from regulators watching how AI handles personal data.