There is a number in the Urban Science case studies that should have started a small fire in every automotive marketing department in North America, and instead it produced roughly the same reaction as a warranty pamphlet.

Two thirds.

Stop optimizing your programmatic buy toward clicks, retrain the same system on verified vehicle sales, and your cost per sale drops by two thirds. Same platform. Same inventory. Same media team. You just told the machine what you actually wanted, which apparently nobody had thought to do.

I had Carl Matter, Director of Ad Tech Performance at Urban Science, on the ADOTAT Show this week. Urban Science is the consultancy that sits on daily sales feeds from the manufacturers, which means it knows what cars sold, where, and to roughly whom, every single day. Matter has been there ten years, which in agency-adjacent time is somewhere between a papal reign and a geological era.

I went into the interview assuming I was about to hear about incremental progress. I came out of it realizing the industry has been standing perfectly still since roughly the Bush administration while congratulating itself on velocity.

Matter, who is a nicer person than I am, put it through a middle school English class: he quoted Inherit the Wind. Motion is relative. Perhaps it is you who have moved away by standing still.

I would have gone with something ruder. That is why he works at Urban Science and I write this newsletter.

Quick orientation, because automotive has its own dialect.

Tier one is national. Ford is a truck company, Honda is reliable, here is a spot with a drone shot of a canyon. No dealership in sight.

Tier two is regional. Your Metro Detroit Ford Dealers, gathered in a co-op, buying the local news break.

Tier three is the dealership itself. Come see Carl Matter Ford, this incentive ends Sunday, we will not stop calling you.

Here is the thing worth sitting with: tier three collectively spends about as much as tier one. It is just shattered across thousands of dealers, which means nobody experiences it as a single number, which means nobody audits it as a single number.

So I asked the obvious question. Do the national brands track a tier one impression all the way down to a car leaving a tier three lot?

Matter: "Typically they do not."

I pushed. Not a single major brand, top to bottom?

He did the thing that consultants do when the honest answer is embarrassing to somebody who is not in the room. He said it is a question a lot of clients have interest in, and that Urban Science is working with a lot of clients to solve it.

Which is a very polite way of saying no.

Billions of dollars in national advertising, and the industry does not close the loop on whether any of it moves metal. It closes the loop on whether you visited a website. I have visited a lot of websites. I have bought one car.

At an earlier agency stop, Matter's team ran a test on their own client's site. The key conversion action was request a quote. You click it, you say what car you want, and in theory a human tells you what it costs.

Matter had his team fill out the form. Then he counted.

Roughly nineteen separate interactions with the dealer, and they still never got a quote.

Nineteen. They were contacted aggressively, enthusiastically, relentlessly. Just never with a price.

Sit with the accounting here. Every one of those form fills registered as a conversion. Every one of them made a campaign look brilliant. Every one of them fed back into an optimization loop that went out and found more people likely to fill out a form and receive nineteen phone calls containing no information.

The metric was not measuring intent. The metric was measuring the industry's own ability to generate voicemail.

This is the part where I told Carl his opinion was crazy, and he agreed, and then made it worse.

A CMO sets a strategy. It is discussed at length, in decks, with consultants, at an offsite. It descends through the agency like a rumor through a high school. And at the bottom, the person who actually executes it, who tags the ad and places the buy and hits the button that spends the money, is entry level. Ad trafficking is the first job you get out of college.

So the strategy is set by the most experienced people in the building and executed by the least, with a game of telephone in between that strips out every trace of why.

Matter's framing was gentle. He said it takes a long time for people to learn to ask the question why, himself included, looking back.

My friend Jordan Finger, who has done about a billion dollars in direct response and was on this show last week, is less gentle. He says the agency model is broken because of junior talent, no accountability, and zero skin in the outcome.

And before automotive gets defensive: I have heard this identical complaint from CPG, from finance, from pharma. Auto is not uniquely broken. Auto is just uniquely measurable, which makes the breakage visible.

Matter added the part that actually explains it, and I will come back to this in part two because it is the whole ballgame: the agency contract itself is usually written around cost per action. Not sales. Actions. The agency is paid to deliver the wrong thing, competently, on time, at a good rate.

You cannot out-strategy your own compensation structure. Nobody ever has.

I asked Matter how much automotive spend goes to campaigns that look busy and sell nothing.

He started with the Wanamaker chestnut, half my advertising is wasted, I just do not know which half. Then he did something more interesting with it.

Of the half that is not wasted, meaning right channels, right macro strategy, roughly half of that could still be substantially better. Call it a quarter of everything, sitting on top of the pile everyone already concedes is garbage.

On a base this size, a single percentage point is real money. He said "more than a lot of us would care to admit," which is consultant for do not make me put a number on the record.

Here is the line that will get Matter yelled at in a QBR somewhere.

If you build your targeting from actual buyers rather than from proxies, your audience gets smaller. Smaller audiences cost more. So your CPM goes from something like four dollars to something like fifteen.

Every media buyer reading that just flinched, because CPM is the number on the scorecard, and the scorecard is the bonus.

Matter's point is that a four dollar CPM against people who will never buy a car is an infinitely bad deal, and a fifteen dollar CPM against people who buy cars is a bargain, and the only reason this is controversial is that one of those numbers is easy to report to a client and the other requires explaining something.

Roughly five percent of the US population buys a car in a given year. About 56,000 sales a day. Fifteen to sixteen million a year, down from the good years. You are hunting a very specific animal with a very wide net and then grading yourself on the size of the net.

Everything above is the funny part. Nineteen phone calls and no price. Twenty two year olds spending national budgets. A two thirds cost reduction sitting on a shelf like a lawnmower nobody returned.

Part two is the part somebody's lawyer is going to read twice.

Because I found the mechanism. It is not a technology gap and it is not a talent gap. It is one paragraph, in one contract, that makes doing the right thing a breach of your own performance terms. Matter said it out loud, first person, about his own agency days, and I do not think he fully clocked what he had handed me until it was already recorded.

Here is what is waiting for ADOTAT+ subscribers:

The paragraph itself, why nobody has opened it since 2019, who actually wrote it, and why every CMO complaining about agency accountability is complaining about a document their own procurement team drafted.

The math that punishes competence. The agency that delivers the same cars for a third of the budget just cut its own revenue by sixty seven percent. Now run that meeting in the back half of a bad quarter.

Five questions I would ask Urban Science before believing a word of the two thirds, including the one about control conditions and the one about the word "permissible," which is not a verbal tic, it is counsel.

The window problem, which quietly invalidates a filing cabinet of reports already delivered to clients who paid for them. Thirty to ninety day measurement against a six to eighteen month decision. The error is not noise. It has a direction, and it is why you keep concluding that retargeting is your best channel.

Why the tier one number does not exist, which is my inference and not Matter's, and which involves four different parties who can each say truthfully that they did not decide this.

Going dark, the only test in advertising that produces an answer nobody can spin, and the reason your agency will never propose it.

And the boy in the hospital, which is where the whole thing actually started, and where it ends.

This is what subscriptions buy. No sponsor saw a draft. Urban Science did not see a draft, including the sections taking their case study apart, and they are a source I like. No holding company gets a courtesy call before a hard piece runs.

Nobody funds this except you, which is the entire reason it can say any of it.

Subscribe. It is cheaper than the DSP fees you are not auditing.