Douwe Bergsma walked into our meeting with the kind of opening line that only a CMO who knows exactly what he's doing would deliver.
"You may be wondering why a company that sells toilet paper, paper towels, and napkins needs a data management platform." He paused, just long enough to let the absurdity land. "We thought perhaps you could tell us."
I've been in hundreds of conference rooms over twenty years — pitching DMPs, then CDPs, now agentic infrastructure — and Douwe's question remains the most honest articulation of the CPG data problem I've ever heard. Georgia-Pacific makes Brawny, Quilted Northern, Dixie, and Angel Soft. Hundreds of millions of Americans buy these products. Georgia-Pacific has no idea who any of them are.
That's not an exaggeration. It's the structural reality of consumer packaged goods. When someone buys Brawny paper towels at Walmart, Walmart knows. Georgia-Pacific gets an aggregate purchase order. The person who chose Brawny over Bounty — the actual human whose preference determines billions in market share — is invisible to the company that made the product.
In 2017, the answer was behavioral data capture. Georgia-Pacific spent millions on digital media, and every ad impression, click, and site visit generated signals about the people engaging with their brands. The DMP could help them organize those signals into something actionable — audience segments for targeting, lookalike models for prospecting, attribution data for optimization. They couldn't build a CRM, but they could build a behavioral proxy for customer understanding.
It was clever. It was expensive. And it's being made irrelevant by the same force that's reshaping every other story in this series.
The Invisible Customer, Revisited
Georgia-Pacific's problem was never really about data management. It was about a structural gap in the value chain: the people who manufacture the products and the people who consume them are separated by retailers who control the transaction and own the relationship. DMPs, CDPs, and every other acronym the industry has produced were attempts to build a bridge across that gap — to give brands some visibility into the customers they couldn't directly reach.
The agentic era doesn't close that gap. It blows up the entire topology.
When a customer's agent manages the household's consumables — paper towels, toilet paper, napkins, cleaning supplies — it doesn't browse the paper towel aisle at Walmart. It doesn't see the end cap display. It doesn't notice the $1-off coupon. It evaluates structured product data against the household's preferences and constraints, and it places an order through whatever fulfillment channel offers the best combination of price, speed, and reliability.
In this model, the customer isn't invisible to Georgia-Pacific because a retailer sits in between. The customer is invisible because their agent is doing the choosing, and Georgia-Pacific's products aren't represented in the agent's decision architecture.
The first challenge was knowing the customer. The second is being known by the customer's agent. Same company, same products, completely different problem.
The Commodity Trap
Here's where it gets genuinely scary for CPG companies. Paper towels are, from a functional standpoint, a commodity. They absorb liquid. Some are more absorbent than others. Some have select-a-size sheets. Some are recycled. But the functional differences are marginal, and the price range is narrow.
Humans choose between Brawny and Bounty for all kinds of reasons that aren't rational: brand familiarity, the lumberjack on the packaging, a childhood association, a vague sense that one feels "tougher." These are exactly the kinds of factors that sixty years of brand marketing have been designed to cultivate.
An agent doesn't have a childhood association with Brawny. An agent evaluates the attributes that its principal has expressed as preferences: absorbency rating, sheet count, price per sheet, sustainability certifications, delivery availability. If Brawny and the store brand are functionally equivalent and the store brand is $2 cheaper, the agent chooses the store brand. Every time.
This is the commodity trap, and it's the existential threat for every CPG brand in the agentic era. When the decision-maker is an algorithm optimizing on expressed preferences, and the product category has minimal functional differentiation, brand premium evaporates. The agent doesn't pay extra for the lumberjack.
Escaping the Trap
So how does Georgia-Pacific — or any CPG company — avoid being commoditized into irrelevance by their customers' agents?
The answer, I think, involves three moves that most CPG marketers haven't started making.
First: make brand attributes machine-readable. If sustainability matters to a growing segment of consumers — and it does — then Georgia-Pacific's sustainability certifications, recycled content percentages, carbon footprint data, and supply chain transparency metrics need to be available as structured data that an agent can query and compare. Not buried in a corporate responsibility PDF. Not on a webpage optimized for humans. In an API that an agent can evaluate in milliseconds alongside every competitor's equivalent data.
Second: invest in the attributes that agents weight. This is where product development and marketing strategy converge. If agents are optimizing on absorbency per dollar, sheet count, and sustainability scores, then those are the dimensions where you need to win. Not win in a focus group. Win in a structured comparison that an agent runs across every product in the category simultaneously. The product roadmap needs to be informed by what agents evaluate, not just what humans perceive.
Third — and this is the most counterintuitive move — double down on human brand-building, but do it earlier in the funnel. Remember: agents don't set preferences. Humans do. The customer who tells their agent "always buy Brawny" encoded that preference somewhere. Maybe it was a childhood memory. Maybe it was an ad that landed at the right moment. Maybe it was a recommendation from a friend. Brand marketing's job isn't to influence the agent — it's to influence the human before they set the agent's parameters.
This means brand marketing and agent-readiness aren't competing priorities. They're sequential layers of the same strategy. Build the brand in the human's mind. Then make sure the brand is discoverable and competitive in the agent's evaluation set. If you do the first without the second, the customer says "I like Brawny" but their agent buys the cheaper alternative. If you do the second without the first, you're competing on pure commodity terms with no preference advantage.
Douwe's Question, Redux
If Douwe Bergsma walked into my office today and asked the same question — why does a toilet paper company need [whatever we're calling the agentic equivalent of a DMP]? — I'd have an answer that's both simpler and harder than the one we gave in 2017.
Simpler: because your customers' agents are going to decide what paper towels to buy, and right now they can't find you.
Harder: because making a CPG brand agent-discoverable requires rethinking how you represent your products at the data layer, how you differentiate on attributes agents actually weight, and how you coordinate brand marketing with agent-readiness in a way that nobody in the industry has figured out yet.
The DMP was a $500,000-a-year solution to a problem of behavioral signal capture. The agentic challenge is a fundamental rearchitecture of how the brand exists in the decision ecosystem. It's not a technology purchase. It's an organizational transformation.
Douwe had the right instinct in 2017. Georgia-Pacific needed to understand their customers, even though they couldn't touch them directly. The mechanism has changed — from capturing signals about anonymous users to being present in agent decision architectures — but the underlying imperative is identical.
You sell toilet paper. Millions of people need it. And increasingly, their agents are deciding which brand to buy. Can you tell me what that's worth?