AI is everywhere. From creative to copywriting, audience insights, segmentation, planning, buying, activation, personalization, optimization, and even measurement, there’s no corner of the marketing world that AI hasn’t reached. A whopping 73% of marketing teams today use genAI regularly, and analysts rank it as the top marketing trend among global marketers in 2025.

There’s a lot of excitement, and rightly so. But for all the opportunities and capabilities it will unlock, we need to be thoughtful and intentional about how we approach a future where every company has its own AI solutions and models. There will be a clash of AI, so to speak, as brands, via agencies, leverage homegrown AI models to make sense of their marketing while competing with the AI models of individual media owners.

There’s a real risk of going down a path of further isolation, confusion, and secrecy as large media owners and social platforms leverage AI for audience activation and measurement. That’s not sustainable. At InfoSum, we believe that the only way to capitalize on AI’s incredible potential is by being fully transparent on the data used to train models, and a significant part of that transparency comes from a strong commitment to measurement. 

AI-enabled advertising is now

Consider the announcements at the recent upfronts and newfronts in the US:

• FOX launched an AI-driven converged media platform to leverage audience and contextual signals and streamline end-to-end operations for its clients across Fox Entertainment, Fox Sports, Fox News Media, and Tubi.
• Netflix is closing in on 100 million global monthly active users on its ad-supported plans and presented a revamped ad suite with new GenAI-based ad formats, expanded data partnerships, and brand lift capabilities.
• Amazon announced richer audience targeting, AI-based contextually relevant pause ads, and new shoppable ad formats for brands that are both endemic and non-endemic to the platform.
• Disney announced an expanded first-party data hub and new AI capabilities, enabling brands to test and refine their creatives on the fly.

Add to that list what the top social media companies, commerce media networks (like Amazon and Walmart), and other key CTV players (like DIRECTV or Samsung TV) are doing to develop rich, new ad inventories and streamline programmatic advertising on their platforms, and the need for measurement standards, or at the very least unified measurement solutions, is more pressing than ever.

Measurement is more than a dashboard

Measurement involves some of the most sensitive consumer data around: transactions, conversions, locations, media exposures, and even very personal demographic and financial data. Advertisers want to know how their campaigns perform across their entire media mix, with enough granularity to allow them to make quick tactical adjustments. However, most media companies today are very skittish about sharing internal data.

So, to err on the side of caution, most large media companies today only share aggregated data with advertising partners, if they share anything at all. More often than not, advertisers are asked to trust top results on a dashboard with limited insights into how specific creatives, placements, audiences, and tactics performed. And they can’t feed those results into AI models and expect them to outperform their previous iterations.

But there are ways to learn from everyone’s granular data without exposing anyone’s privacy. Let me walk you through a few privacy-enhancing techniques (PETs) that we’ve developed at InfoSum, which form the foundation of our privacy-by-default approach to collaboration. 

PETs safeguard critical measurement processes

Our data collaboration solutions at InfoSum are all based on a core set of principles:

• Data isolation: We’ve designed a system where each party’s raw data is stripped of any personally identifiable information (PII), transformed into a mathematical representation, and contained in a Bunker that’s entirely under its owner’s control. No raw data or PII changes hands at any point in time.
• Access controls: Each party has granular permissions and access controls allowing it to define, monitor, and control what actions can be performed on its dataset, for how long, and at what level of granularity. 
• Decentralized processing: All data processing takes place in a decentralized environment. Queries are broken up and reconciled across edge devices without sharing any of the input data or moving it around.
• Data obfuscation: We employ techniques such as differential privacy and privacy budgeting to add noise to query results and limit the number of queries any one party can run within a 24-hour period, thereby preventing reidentification through brute force. In addition, rounding and redaction thresholds give users control over the size of their aggregated results, providing the ability to create custom cohorts in the platform.
• Transparency: Most data matching and modeling today occurs in a black box, without visibility into how data is collected, used, transformed, or shared. Not with us. We provide companies with complete transparency into how their data is used to fulfill specific queries.

With those precautions firmly in place, advertisers, agencies, and their media partners can feel confident that their data is safe and that they have complete control over how it’s being used. They can connect their data assets to measure cross-platform reach and frequency, analyze audience duplication, or run actual science-based incrementality tests without compromising the underlying data.

A new role for agencies

AI is radically changing the way agencies work, and they’re now in a unique position to orchestrate data collaboration between their clients and media partners. The large media holdcos in particular have a lot of leverage to develop comprehensive marketing solutions for their clients across all the top publishing platforms. Brands want to use granular data at scale to make quick, smart, and cost-effective decisions with their campaigns, and they don't have the time or resources to make sense of dozens of independent data sources and KPIs. 

They're asking agencies to help them connect the dots. To achieve this, privacy protection is the only viable option. With decentralized data collaboration and fully controllable privacy safeguards in place, large media owners and walled gardens can begin to lower their drawbridges and become part of the vast, interconnected media ecosystem currently under construction. Bilateral understanding allows their data to be used safely by agency AI models, and vice versa, with brands requiring transparency and privacy protections when large platforms leverage their data.

If you want to learn more about how to translate better measurement into better intelligence, and better intelligence into winning marketing decisions, get in touch with the team.