The 2022 presentations to CPG were gorgeous. Vendor decks with gradient boxes around a flywheel diagram that promised with the serene confidence of a Greenwich Village psychic that first-party data maturity was three strategic phases and one enterprise CDP away. Media and digital leadership ate it all up and became true believers because the alternative was having to explain to their CMO that Google was threatening again to remove cookies (spoiler alert: they never did), Apple had kneecapped mobile tracking infrastructure, signals were going dark, and the plan was to wait and see. Nobody makes VP waiting and seeing.
So the CDPs got purchased. The data teams got hired. The loyalty apps got launched. QR codes went onto tens of millions of units of product like little paper lottery tickets nobody would scratch. The clean room partnerships got signed. And internal decks about "the journey toward first-party data maturity" got presented to people who wouldn’t ask hard questions because they didn’t want hard answers.
Four years on and retail media's rented data bill is higher than it's ever been and few in CPG offices find this disorienting.
Start at the top with P&G and their rare gift for institutional honesty. In 2020, the most powerful consumer goods company on earth published an innovation brief on MIT's startup exchange asking for help figuring out how to collect first-party data at scale. Their existing data had arrived "in small bites" through sweepstakes and on-pack QR codes. They couldn't crack the value exchange and shared this publicly, in writing, for the whole industry to read.
Nothing in the public record suggests they found the answer. So they pivoted themselves to owning something else and with a wink called it consumer data. But it’s really not.
The ad tech contingent waves at P&G's 1.5 billion consumer IDs as vindication. Look harder though and you’ll realize that what P&G assembled was an anonymous device ID graph built from open-web programmatic bidstreams and retailer loyalty partnerships. A device ID is not a consumer relationship though. At best it’s a pseudonym for a mobile device. P&G couldn’t and didn’t convince 1.5 billion people to hand over their email addresses so they built a sophisticated machine to bypass platform walled gardens and manage ad frequency. .
A well-done exercise in efficiency, but in reality they just papered over the quest for owned consumer data problem with better engineering.
Colgate-Palmolive told AdExchanger in 2024 they'd "always had to trade on data we don't own or control" and that the industry had "gotten used to being dependent on other people's data." Their solution, after covering their brand site with email capture requests, was to route those consumers to omnichannel retailers; concluding the path to data freedom ran back through the people they were trying to escape.
Defenders will highlight Hill's Pet Nutrition, and fair enough. Colgate's premium pet food business grew D2C subscriptions by double digits and secured real first-party data. You know why? Because a consumer buying a $90 bag of prescription kidney-care kibble for a fifteen-year-old cat is extremely motivated. The lifetime value is enormous and the consumer will register, subscribe, and check in monthly because the alternative might be their cat dying. Take that same infrastructure, that headcount, that vendor tax, and apply it to a $4.96 tube of Colgate Total and you've built a money furnace with PR window dressing.
General Mills made the most noise when they trumpeted their Good Rewards program with the Fetch app that got two million members in its first year. The catch was that it flowed through Fetch’s third-party app built on uploaded photos of retailer receipts. Still renters believing they're landlords.
Unilever spent years boasting about owning almost one billion consumer records accumulated across dozens of regional digital marketing hubs. Then new leadership arrived in 2025 during the post-inflation margin squeeze, looked at the multi-million dollar annual maintenance bills, and asked the basic question that should have been asked in year two: did this actually sell more soap? Their answer was to pivot to a few hundred thousand influencers to drive "desire at scale" instead; a different problem to litigate another time.
The reason none of this works at scale for CPG is structural, and every serious operator in the space already knows it. First-party data matters when the consumer must identify themselves to complete a transaction. Banks know who you are because anonymous money transfer is a federal crime. Nespresso knows who you are because the pod subscription tied to a proprietary registered machine needs a shipping address. The data is a consequence of the business model not a strategy layered on top of it.
CPG sells through retailers. The consumer hands the box to a cashier and walks to the parking lot. No loyalty app repairs that architecture because retail’s architecture is load-bearing.
And yet the martech stacks tied to hallucinations of owned consumer data kept growing because institutional complexity is self-perpetuating. The CDP requires an identity layer, then the identity layer requires a clean room, and then the clean room requires dedicated analysts. Each layer is technically necessary to the layer beside it, and none of them require anyone to ask whether the whole structure produces something a smaller, cheaper version couldn't.
The SaaS vendors knew this when they wrote the pitch decks. You don't build a multi-hundred-billion-dollar global ad tech market by solving problems cleanly. The average cost of operating a single data clean room is $376,000 before you even get to the CDP underneath it or the cloud warehouse overages. The infrastructure sold to liberate CPG from data dependency is being enabled (and sometimes sold) by the same data platforms and retail media networks CPG was supposedly escaping. Both parties charging CPG rent on an escape tunnel fantasy that leads right back to the cell.
Like so many advertising industrial complex KPIs, the KPIs measuring first-party data maturity were input metrics, every last one. Records in the database. Cost per acquisition. CDP utilization rate. Connected packs live across markets. Nobody bothered to build a dashboard to track whether any of it shifted a dollar of retail media spend toward owned channels, reduced dependence on rented audiences, or changed the negotiating position with a single retailer. All the while, the empire of owned data was, in practice, an expensive way to feel strategic.
The thin version of this approach does produce something real at the margins. A small first-party seed audience fed into a lookalike model produces measurable expansion of targets and a real lift. But you don't need an enterprise-grade seven-figure CDP stack to hold 100,000 superfans. An Excel export can handle it.
The useful version of this strategy is less romantic and considerably cheaper. Be a smart, honest, and proud renter. Spread across landlords so no single one can squeeze you unilaterally. Use whatever data you have to sharpen the audiences you're buying access to, accept that the retailer will always have a better purchase signal than you do, and negotiate accordingly. The shelf is theirs. The data is theirs. That's the deal. In return, you get the brand which is worth more than any of it if. That's what you've been neglecting.
*** Views expressed on Push Pull Pivot are personal and shared in an individual capacity. They do not represent those of any current or former employer ***