Rethinking Media Strategy Beyond the Click with AI, First-Party Data, and Customer Segmentation

Interview with Kip Pogrebenko | CEO and Co-Founder, Medloaner | Former data science lead at Omnicom Media Group | Worked on Amazon Marketing Cloud at Amazon

Published: July 3, 2025 · Updated to correct a statistic about iOS 14.5, clarify the expert's role at Amazon and add privacy considerations.

Disclosure: The expert is co-founder of Medloaner and of Red Flag, both mentioned in this article.

When Apple let iPhone users refuse cross-app tracking in 2021, most of them did, and much of digital advertising lost its targeting almost overnight. Kip Pogrebenko, who has worked on data products at Omnicom Media Group and on Amazon Marketing Cloud, took a simple lesson from it: build on data you own, segment before you create, and treat governance as part of the product.

Own Your Data, or Rent Someone Else's Risk

Pogrebenko points to Apple's iOS 14.5 release in April 2021, which introduced App Tracking Transparency and required apps to ask users for permission before tracking them across other companies' apps and websites. Most users declined. "There was almost an apocalypse," he recalls. Advertisers that depended on cross-app tracking lost much of their targeting and measurement, a reminder that strategies built on data you do not own can collapse without warning.

First-party data gives companies control over how data is processed, modeled and activated. It becomes more powerful inside clean rooms such as Amazon Marketing Cloud (AMC), Amazon Ads' clean-room analytics product. Brands can bring their own first-party data, such as customers acquired through paid search, and see in aggregated form how those audiences behave across Amazon's properties. Amazon's lookalike modeling can then extend reach to similar shoppers. "When people come to Amazon.com, they know what they want to buy," he says.

Privacy Comes With the Territory

Owning customer data also means owning the obligations that come with it. Clean rooms are designed so that advertisers and platforms can match and analyze audiences without exposing individual-level records to each other, typically using hashed identifiers and minimum aggregation thresholds. That does not remove the need for a lawful basis: first-party data used for advertising must have been collected with clear notice and, where required, consent, under rules such as the EU's GDPR and the California Consumer Privacy Act. A single, well-governed source of truth, the principle Pogrebenko applies to AI, matters as much for privacy as for performance.

Segment First, Build Second

Early in his career, Pogrebenko built a technically sound product that failed because it did not solve a clear customer problem. Studying strategy at Harvard Business School gave him the missing piece: products should be built around segments, not features.

He now launches small. "You launch a product to conquer the smallest segment that you have, but also to get those learnings."

The same logic drives pricing. Some customers will pay $100 a month for one specific feature, while others will only pay $20 for a full suite. "Without understanding this, you'd probably shove that customer into the $20-per-month segment and leave revenue on the table." He uses the Jobs to Be Done framework to score how important each customer problem is and how well current solutions solve it, then prioritizes features accordingly.

Clarity Beats Creativity, Then Creativity Wins

"Creativity comes second," Pogrebenko says. Clarity and segmentation come first. He once tested a bare message, "a white PowerPoint slide with just three words," that outperformed more polished content because it named a known pain point.

B2B is changing, though. Younger decision-makers raised on TikTok, YouTube and Instagram expect consumer-grade design and storytelling. "We call it the gray carpet office world, and it's dying." Mid-market buyers such as auto dealerships increasingly need consumer tactics. "You have to do B2C strategy because you can find them on streaming TV, on YouTube."

One Source of Truth Before Any AI

"There has to be a single source of truth," Pogrebenko insists. Without it, metrics conflict, teams work in silos and troubleshooting becomes guesswork.

With that foundation, his team built and launched a new app, Red Flag, in seven days using internal AI agents. "We built it for ourselves, and nevertheless we went from scratch to launch in seven days."

AI still makes mistakes with brand consequences. In one case, an AI-generated video announced the "launch of Safe Flag" instead of "Red Flag." Small errors like that can erode trust, and in regulated companies every creative still needs legal and compliance review. "All messaging in creatives has to go through legal and compliance."

Pogrebenko's Three Imperatives

• Anchor every initiative in customer segmentation: not only marketing, but pricing, product and messaging.
• Treat first-party data as a core business asset: for personalization, attribution and strategy, with privacy built in.
• Prioritize team alignment and data consistency: to reduce friction and scale execution.

He compares a strong team to a jigsaw puzzle. "Everyone has its own shape, everyone has its own complementary responsibilities. There should be no duplication of efforts."

Kip Pogrebenko is the CEO and co-founder of Medloaner, which builds AI tools for hospital supply chains. He previously led data science work at Omnicom Media Group and worked on Amazon Marketing Cloud at Amazon.

Sources

• Apple, App Tracking Transparency (iOS 14.5, April 2021)
• Amazon Ads, Amazon Marketing Cloud overview
• GDPR (Regulation (EU) 2016/679) and California Consumer Privacy Act

Related on the Community Hub

• The Rise of First-Party Data and Clean Rooms
• Driving SaaS Growth Through Product Usage Signals
• The Next Wave of Streaming: Precision, Partnerships, and Personalization

‍