Retail Media. Commerce Media. Financial Media. The $200B Shift Third-Party Cookies Made Inevitable.
A practitioner’s guide to the emerging marketing, advertising, & data (MAD) tech infrastructure reshaping consumer industry: what it is, who’s winning, and which careers it’s creating right now
You’ve been hearing the acronyms for two years. RMN. CMN. FMN. Retail media. Commerce media. First-party audiences. First-party data moats.
If you’ve been nodding along in meetings this post is for you. Plain English. Real brand stories. What AI is honestly doing to it. And what it means for your career.
Start Here: What Problem Does This Actually Solve?
For twenty years, digital advertising ran on third-party cookies tiny trackers that followed you around the internet. Then Apple’s iOS 14 let people opt out. Privacy laws tightened. Google spent five years threatening to kill cookies then backed down. But the ecosystem had already moved on.
Here’s what most people missed: it doesn’t matter. First-party data what you actually own about your customers or audience is now the required infrastructure for performance advertising. That direction is not reversing.
The question the industry answered: if you can’t follow people around the internet, how do you target them? Own the transaction data closest to purchase. What someone bought, where, when, with what basket and across which merchants. That data lives inside three kinds of companies. That’s what RMN, CMN, and FMN are.
The Three Layers
RMN (Retail Media Networks) — The Foundation. Retailers sell ads using their own purchase data. Amazon ($56B in ad revenue, third-largest ad platform globally), Walmart Connect, Target Roundel, Kroger there are now 277 retail media networks worldwide as of November 2025. Up from a handful five years ago.
The hard truth: in the US, Amazon and Walmart will capture 89% of incremental retail media spending in 2026 meaning every other network competes for the remaining 11%. If you’re a brand trying to “diversify away from Amazon,” the math tells you why that strategy feels hollow.
CMN (Commerce Media Networks) — The Expansion. Any business with transaction data can build an ad business. Instacart alone hit $960M in US ad revenue in 2025 — and its OpenAI integration makes it the checkout layer inside ChatGPT, the first grocery app to offer end-to-end shopping and payment without leaving the chat. Marriott launched a media network in June 2025 — 237 million Bonvoy members, pilots with United, Uber, Visa, Starbucks. United Airlines offers an average of 3.5 hours of captive attention per journey. DoorDash, Gopuff, Lyft every order is a signal, every delivery is an ad opportunity. McKinsey projects commerce media growing at 21%+ CAGR through 2027, faster than display, CTV, or search.
FMN (Financial Media Networks) — The Highest Signal Layer. Retailers see what you bought inside their walls. Banks see what you bought everywhere. That is a different class of signal.
The distinction from CMN isn’t the industry - it’s the data. Financial media networks have visibility across all your merchants, not just their own ecosystem.
Chase’s cross-merchant spending portrait groceries at Whole Foods, running shoes at Nike, hotel in Miami is something no retailer can replicate. Chase Media Solutions launched in April 2024 (80 million customers). PayPal Ads Manager launched October 2025 (400 million accounts, nearly $500B in annual transaction volume, no upfront costs democratizing high-signal targeting for SMBs). Klarna: 100 million users, 724,000 merchants.
US FMN ad spending grew 83% year-over-year in 2025, reaching $640 million at a 107% CAGR through 2026. Citi, Capital One, Amex, Mastercard are next. Almost nobody has this layer in their media plan yet. That gap is the opportunity.
What AI Is Actually Doing — And Where the Real Magic and Moat Is Being Built
Most brands still talk about AI in commerce media as if the prize is better bidding, faster creative, or smarter targeting. That’s the visible layer. It matters. But it’s not the moat and it’s not the magic either.
The real magic is simpler and harder: brands learning who their audience actually is. Their taste. Their timing. Their triggers. Not as a segment or a persona deck, but as a living, compounding understanding that deepens with every interaction.
That’s what the decision loop is really about. Not automation. Relationship at scale.
That’s the divide. Who controls the data, the identity, the activation, the learning loop and the memory of what worked and why. That’s what separates brands that build genuine audience relationships from brands that just rent attention. That divide is opening up right now, and most teams don’t know which side they’re on.
There’s also a delicate tension worth naming: monetizing audience data through media is rewarding when done well, but corrosive when done carelessly. The brands winning long-term are the ones treating audience intelligence as a relationship asset, not a revenue lever. The ones losing are the ones optimizing impressions at the expense of trust. Here’s what the winning side of that equation actually looks like:
The data cloud is the new center of gravity. Snowflake, BigQuery, Databricks this is where first-party data lives, and the brands treating it as the one true source have a real advantage over brands renting that capability through a vendor’s black box. Every vendor-centric approach creates the same trap: you scale fast, lose control of the logic, and can’t port your learnings when the contract ends.
Advanced Identity resolution is the first moat. Most brands are targeting a loud, easily identifiable subset of their own customers logged-in, loyalty-enrolled, already buying. High-value prospects who haven’t touched your ecosystem are invisible. Real ID resolution stitches loyalty IDs, hashed emails, POS transactions, app behavior, and clean-room-matched partner data into a single person not a device proxy. The brands who’ve built this graph are reaching audiences their competitors literally cannot see.
Activation and matching is where data becomes money. Unified identity means nothing if it stays in the warehouse. Activation, pushing enriched, identity-resolved audiences to DSPs, ad servers, retail networks, and CTV simultaneously is the execution layer most teams are still doing manually or not at all. The gap between “we have the data” and “we activate it in real time across every surface” is where most brands are losing.
Decision traces are the emerging moat nobody’s talking about yet. Most teams keep reports. Very few keep a usable memory of why a decision was made, what signal supported it, and whether it actually worked. The ones building decision traces structured records linking Intent → Evidence → Action → Outcome → Learning are accumulating an asset their competitors can’t replicate. Each cycle makes the next bid smarter, the next audience tighter, the next creative more precise.
This is the real meaning of “AI-powered” in commerce media. Not a better creative generator. A decision loop where every dollar spent feeds back into better decisions on the next dollar. The networks and brands that close this loop compound. The ones running disconnected point solutions don’t.
Most brands are still at table stakes automated bidding, keyword harvesting, basic AI creative. The infrastructure layer underneath data cloud, identity graph, activation pipeline, decision traces that’s not aspirational. That’s being built right now by the teams that will own this category in three years.
That decision loop matters most where the purchase actually happens and AI is now disrupting which surfaces those are.
The Unexpected Winner: In-Store
AI shopping agents disrupt the digital shelf if an agent browses and buys for you, you never see a sponsored listing. The one surface where the final purchase decision still belongs to a human, not an AI agent browsing on their behalf: the store aisle.
Here’s the gap that explains why this matters more than most realize: 80% of consumer spending happens in-store. 90% of retail media advertising is online. That mismatch is the white space.
Retailer apps now have full “store mode” store maps, aisle navigation, personalized offers, scan-and-go checkout, AI chat. The most significant in-store activation in 2026 isn’t the screen in the parking lot. It’s the phone in the shopper’s pocket.
The most underreported data play: Walmart acquired Vizio in December 2024. Vizio TVs are being transitioned to sell exclusively at Walmart and Sam’s Club, with potential to reach 25–30% of US households through Vizio OS and its Onn TV brand TV viewing linked to purchase across more than 4,600 US store locations. The timing matters: TVs are purchased every 6–7 years, and 2026 is the replacement cycle. Add the World Cup (104 matches, cross-device streaming, ad inventory already 90% sold out on Telemundo/NBCU) and Walmart is about to close the living-room-to-checkout loop in a way nobody has done before.
The Measurement Problem Nobody’s Solved
Retail media is growing faster than its measurement discipline. That gap is costing brands real money.
Here’s the stat that should reframe every planning conversation you have this year: only 15% of brands report strong confidence in their retail media measurement. Yet 86% say strengthening measurement is a high or critical priority. And only 12% have reached true full-funnel capability across on-site, off-site, and in-store.
The industry is simultaneously scaling fast and flying partially blind.
Most reported ROAS is correlation, not causality you served an ad to someone who was already going to buy, and called it a win. The brands quietly building incrementality testing infrastructure right now clean rooms, holdout groups, causal lift measurement will make better budget decisions than everyone else in 18 months and be able to prove it. Meanwhile, 52% of advertisers are already moving display budgets from open-web DSPs to retail media DSPs, driven by the promise of closed-loop attribution. The problem is the promise is ahead of the reality. Most networks still can’t deliver it at full-funnel scale. The budget is moving. The infrastructure to justify it hasn’t caught up.
What It Means for Your Career
The broader marketing job market is healthy but competitive. 65% of marketing leaders plan to expand permanent headcount in H1 2026 but 45% say finding skilled professionals is harder than a year ago. The gap isn’t in headcount. It’s in specialization. 78% of hiring managers now pay premiums for candidates with specialized skills over generalists. The roles below are where that premium is concentrating in commerce media specifically, besides execs who is launching this as GMs.
Roles with real demand right now:
Retail/Commerce Media Manager — managing budgets across Amazon, Walmart, Instacart, Kroger, and the growing FMN layer. LinkedIn consistently shows hundreds of open roles tagged “retail media” alone, with tens of thousands more in commerce media broadly. This title barely existed five years ago and commands a clear premium especially at the senior level at agencies and brands.
Marketing Analytics Manager / Measurement Lead — Incrementality testing, clean rooms, cross-network attribution. Robert Half’s 2026 Salary Guide projects Marketing Analytics Manager among the strongest salary growth trajectories in all of marketing. With only 15% of brands confident in measurement, this skill is both scarce and increasingly non-negotiable for budget justification.
First-Party Data Strategist — auditing what data you own, building the identity resolution roadmap, connecting warehouse infrastructure to activation. The title is still emerging and inconsistently labeled you’ll see it as “Customer Data Platform Manager,” “Identity Strategy Lead,” or “Audience Operations Manager.” The role is real even if the title isn’t standardized yet. 56% of marketing leaders report skills gaps in their departments this is the gap most are struggling to fill.
Decision Architect or Engineer — building and maintaining the data cloud → identity → activation → decision trace pipeline. The most genuinely nascent role on this list. No standard job title yet, but the function someone who owns the technical layer between your warehouse and your media activation is what every sophisticated marketing team needs and almost none have. Watch for it emerging from data engineering, marketing ops, and ad tech backgrounds over the next 18–24 months.
Under pressure: the manual optimization layer - daily bid adjustments, pacing reports, weekly dashboard updates. AI does this in real time. Also under pressure: third-party audience buyers and cookie-based targeting specialists the skill set built for the last era doesn’t transfer cleanly to this one. Strategy, creative judgment, incrementality interpretation still human. Administration and legacy targeting logic are not.
Skills that compound regardless of role: understanding incrementality (not ROAS, but what caused the sale), comfort with data clean rooms, ability to translate between CMO and CDO, familiarity with warehouse-native data architecture, and enough technical literacy to speak the language of a Decision Architect.
The One-Paragraph Summary for Any Meeting
Retail, commerce, and financial media networks are what replaced third-party cookies. Instead of tracking people across the internet, brands now buy access to retailers’, travel companies’, and banks’ first-party transaction data more accurate, more consent-based, more connected to real purchase intent. Amazon built the template; the model spread to 277 networks globally; banks are entering with structurally superior data; AI is simultaneously making these networks more powerful and disrupting the on-site formats at their core; and Walmart is linking TV viewing to store purchase at a scale nobody has done before. The brands and marketers who understand how to operate in this infrastructure not just spend in it will pull away from everyone else.
This is no longer a channel story. It’s a control story.
Who owns identity. Who controls activation. Who keeps the learning. Who can prove causality instead of reporting correlation.
That’s where the advantage will sit.
Retail media was the first chapter. Commerce media is the expansion. Financial media may become the highest-signal layer of all.
Most teams still see inventory. The smarter ones see infrastructure. The winners will see decision systems.
Next week: Part 2 — The Infrastructure War
Now that you know the landscape, the real question is: who actually wins it?
Not the network with the most inventory. Not the brand with the biggest budget. The winner is the company that built something competitors can’t buy a brand customers seek out, or infrastructure that compounds with every dollar spent.
Most companies have neither. They’re renting both.
Part 2 covers the decision infrastructure underneath the campaigns the Renters vs. the Builders, the identity tax most teams don’t know they’re paying, and the five questions every CMO, CDO, and GM should ask before allocating another dollar.
Author’s note: An LLM helped with research, citations, and light proofreading without changing content. All ideas, arguments, voice, and em dashes are mine. :)