In 2018, Joe bought a car. Let me tell you what that looked like from the data side, because it's the best illustration I've ever found of how broken the customer journey really was.
It started with a cable TV ad — untrackable. Then a Facebook ad that caught his attention — tracked, but siloed. He browsed inventory on his tablet during lunch — different device, different cookie. Switched to his desktop at home for deeper research — another session, another identity. Applied for credit pre-approval on his phone — yet another system. Visited a dealership that weekend — in-store beacon detected his phone, but that data went nowhere useful. Visited a second dealer the following weekend. Signed up for email alerts from both. Got retargeted on YouTube with video ads for the car he'd already test-driven.
Hundreds of signals across dozens of systems. Nobody saw the whole picture. The cable company saw one thing. Facebook saw another. The dealers each saw their fragment. The credit bureau had its own view. The retargeting platform kept serving ads for a car Joe had already decided to buy.
For two years after buying his car, Joe got served ads for the car he already owned.
This was the customer journey in the pre-agentic era: a fragmented mess of signals, identities, and systems that couldn't talk to each other. The entire customer data platform industry was built to stitch this together — to give brands a unified view of Joe's journey from awareness to purchase. It was a noble effort, and it partially worked, but it never fully solved the fragmentation problem because the journey itself was chaotic, multi-device, multi-channel, and fundamentally human.
Joe's Agent Buys a Car
In 2026, Joe needs a new car. Here's what happens.
Joe tells his agent: "I need a mid-size SUV. Budget up to $52,000. Must have adaptive cruise control and a panoramic roof. Good safety ratings — I have a teenager starting to drive. I want to keep my monthly payment under $750. I don't care about brand, but I'm not interested in anything with reliability scores below average."
That conversation takes four minutes.
Joe's agent does the rest. It queries twelve manufacturers' product databases simultaneously. Evaluates specs, safety ratings, predicted reliability, ownership costs, insurance estimates, and resale projections. Cross-references Joe's commute patterns from calendar and navigation data to recommend the optimal drivetrain. Checks Joe's credit profile and pre-qualifies with four lenders in seconds. Identifies three vehicles that match all criteria. Contacts dealerships within a thirty-mile radius, negotiates pricing, and surfaces the best offer.
Joe reviews the agent's recommendation over breakfast. Drives to one dealership. Test-drives one car. Signs the paperwork — financing already approved, price already locked, trade-in already appraised using photos the agent submitted. Drives home in his new car.
Total elapsed time from first conversation to keys in hand: six days. Total touchpoints visible to any single brand: one dealership visit.
The 2018 journey was chaos. The 2026 journey was choreographed by an AI better at car shopping than Joe ever was. The question isn't "how do we track the journey?" anymore. There's no journey to track.
The Death of the Funnel
Joe's 2018 journey was a marketer's dream — if you squinted. All those touchpoints were opportunities for influence. The cable ad created awareness. The Facebook ad built consideration. The desktop research phase was where content marketing did its work. The dealership visit was where the sales team took over. The email nurture tried to keep Joe warm. The retargeting tried to bring him back.
Every stage of the funnel had its own tactics, its own teams, its own budget, its own metrics. The marketing organization was structured around the journey — awareness team, demand gen team, content team, field marketing, digital, each owning their piece of the funnel.
Joe's 2026 journey collapses all of that. There's no awareness stage — the agent doesn't need to become aware of brands; it queries all of them simultaneously. There's no consideration stage — the agent doesn't deliberate over weeks; it evaluates in seconds. There's no nurture stage — you can't drip-email an API. The entire funnel compresses into a single moment of evaluation, and the brand either qualifies or it doesn't.
For automotive marketing organizations — which are among the most funnel-dependent in any industry — this is an existential restructuring. The people, the budgets, the agency relationships, the media plans, the CRM systems — all built for a journey that's disappearing.
What the Dealer Becomes
The dealership model is worth examining specifically, because it illustrates how profoundly the agentic shift changes not just marketing but the entire go-to-market structure.
In 2018, the dealership served multiple functions: showroom (let the customer see and touch the car), information center (answer questions the website couldn't), negotiation venue (haggle on price), and financing desk (originate the loan). The human salesperson was the binding agent across all of these functions.
In 2026, Joe's agent handled information gathering, comparison, negotiation, and financing before Joe walked through the door. The only function the dealership served was the one an agent can't: let Joe sit in the car, feel the steering wheel, check whether his teenager fits comfortably in the back seat.
The dealership becomes a test-drive facility with a signing desk. That's it. All the other functions — the ones that justified the enormous real estate, the sales staff, the F&I department — are being absorbed by agents.
For automotive PMMs, this means the dealership experience — the thing they've spent decades trying to influence and optimize — is being reduced to its irreducible physical core. The marketing challenge isn't "how do we drive foot traffic to the dealership" anymore. It's "how do we win the agent's evaluation before the human ever leaves their kitchen."
The Identity Problem, Solved and Dissolved
Here's the irony that kills me. The entire customer data platform industry — including my own book about it — was fundamentally about solving the identity problem. How do you stitch together Joe's cable impression, his Facebook click, his tablet browse, his desktop research, his dealership visit, his email signup into a single unified profile?
It was hard. Really hard. Cross-device identity resolution, probabilistic matching, deterministic linking, data clean rooms, identity graphs — hundreds of companies and billions of dollars thrown at the problem of "who is this person across all these touchpoints?"
Joe's agent solved the identity problem by eliminating it. There are no fragmented touchpoints to stitch together because the agent manages the entire journey as a single, coherent process. There's no cross-device identity challenge because the agent is device-agnostic. There's no question of "who is this person?" because the agent announces exactly who its principal is and what they want.
The identity resolution industry was the right solution to a problem that's being dissolved rather than solved. You don't need to stitch together a fragmented journey when the journey isn't fragmented anymore.
What Joe Teaches Us
Joe's story — then and now — is the clearest illustration of what changes when the customer delegates. Every complexity that the data industry was built to manage — fragmented signals, cross-device identity, multi-touch attribution, journey orchestration — existed because humans make messy, multi-step, multi-channel decisions. The agent makes clean, fast, comprehensive decisions.
The marketing capabilities that mattered in Joe's 2018 world — awareness, consideration, nurture, retargeting — don't disappear overnight. Plenty of consumers are still buying cars the old way. But the trend is clear: as agents get better at high-consideration purchases, the messy human journey compresses, and the marketing infrastructure built around that journey loses its reason for existing.
For two years, Joe got served ads for the car he already owned. In the agentic future, that particular indignity at least goes away. The agent knows what Joe bought. The question is whether the automaker knows how to talk to the agent — and whether its product data is good enough to win the evaluation that replaces the journey.
Joe's test drive might be his last. Not because he won't buy another car. Because his agent will have done everything else before he even starts the engine.