Which Dollar Sold Which Car?

I. The soup

Most dealer principals I talk to can tell you their closing percentage to within a point or two. They can tell you F&I penetration by product. They can tell you hold per copy, days supply, and allocation by trim level two weeks out. That precision is real, and it's earned.

Ask the same principal which ad dollar sold which car last month, and the answer is almost always the same.

Silence. Then: "We spend about X on digital and Y on radio." Then maybe: "Facebook seems to be working." Then, eventually: "Honestly, we're not sure. The CRM tells us one thing, Facebook tells us another, the agency tells us a third number. It's all kind of a mess."

That's not bad attribution. That's no attribution. They don't have the answer, and — this is the part that matters — most have never sat down and realized they should.

This is the silent gap at the center of most dealerships. The business is measured tightly on the parts you can touch, and left entirely unmeasured on the part that funds the business. Every month the lot turns, the bills get paid, and every marketing dollar flows through channels that nobody in the building can connect to a sold unit.

I'm not calling dealers out for this. Most haven't thought about it because, until recently, they didn't have to. The point of this essay is that "didn't have to" is running out.

II. What dealers actually measure

Before I say what's missing, credit the parts that aren't.

Dealers are obsessive about the numbers that touch the P&L directly. Gross per unit. Packs. Recon cost. F&I penetration by product — warranty, GAP, protection. Closing percentage on showroom traffic. Appointment-to-show ratio. Days-to-sale. Recon turn time. Net net. These get reviewed daily — sometimes multiple times a day — by people whose livelihood depends on knowing them cold.

That discipline is real. Most businesses in most industries do not operate with that kind of granular awareness of their own operations. A franchise dealer who hits their month has earned it.

The gap is one level up from all of that. It's not in the operational numbers. It's in the upstream question: where is the demand that drove those operational numbers coming from?

That question doesn't get asked. Not because dealers aren't smart enough — because there's no clean place to ask it. Marketing sits in one silo. The CRM sits in another. The DMS sits in a third. Each ad platform tells its own story and none of the stories reconcile. Call tracking is its own system. Walk-ins get logged as "Other" by a salesperson who filled it in between walk-ups.

So the question gets dropped. Not answered — dropped.

III. The default isn't wrong attribution. It's abandonment.

Here's what actually happens when a dealer realizes they can't trace a sold unit back to the ad dollar that originated it.

They don't build a better attribution system. They don't hire a data person. They don't rebuild the stack. They do what any rational operator with a P&L to hit would do: they focus on what they can control.

Which means: more leads. More ad spend. More volume through the top of the funnel, because the funnel is the only part that feels steerable. Or — if the month is already past saving — they take the L, blame the market, and start over on the first.

I don't say that to be cynical. I've sat in that room. It's a reasonable response to an impossible measurement problem. You can't fix what you can't see, and you can't see something you've never had a legible picture of. So you stop trying.

But the consequence of "stop trying" is that every month, a dealer is spending a material line item in their budget on faith, habit, and vendor reports. NADA's data puts average dealer advertising spend somewhere around six hundred dollars per new vehicle retailed¹ — real money, spread across channels, sourced from vendors whose reports don't reconcile. Not because the dealer doesn't care. Because nobody has ever handed them the picture that would let them care productively.

When I say "closing the loop," this is what I mean. Not better attribution. Any attribution — deliberate, end to end, from the first ad impression to the signed deal.

IV. Why it's organizational, not technical

The cleanest diagnosis I can give a dealer is this: the problem isn't that your systems don't talk to each other. The problem is that your vendors don't.

Walk through a typical store's stack. The marketing agency runs Meta and Google. A different vendor owns the CRM. A third handles SMS and email follow-up. Call tracking is its own system. The DMS is Reynolds or CDK, managed by another team. The website is a vendor with its own forms, its own chat, its own lead-routing logic. Five or six vendors — sometimes more. Each one reporting the same lead with a different ID, a different timestamp, and a different definition of "conversion."

And then there's the internal side. The salesperson closes a deal and nobody makes them update the source field in the CRM — because the store's incentive is the unit, not the data trail. The BDC sets an appointment and logs it one way; the salesperson codes the customer another way at close. The manager pulls month-end reports that are stitched together across five systems by an office admin reconciling them in a spreadsheet at 10 PM on the 31st.

No single vendor is responsible for the end-to-end picture. No internal seat is either. So the picture doesn't exist.

Fixing this isn't a software problem. A new CRM doesn't fix it. A new attribution tool doesn't fix it. It's an organizational fix: one system, one source of truth, one line of sight from ad platform to sold unit, with the accountability for maintaining that line of sight owned by a named seat in the business.

That's a decision. Not a purchase.

V. The tide is going out

Warren Buffett's line is the one I keep coming back to:

Only when the tide goes out do you discover who's been swimming naked.²

For most of the postwar era of automotive retail, the tide for franchise dealers has been in. Manufacturer demand pushed product through the funnel regardless of marketing precision. Consumer credit was cheap. Inventory shortages helped margins more often than they hurt them. Dealers survived 2008 and COVID because the downturns were temporary and the tools on the other side were the same tools as before — a stronger operator could just wait the cycle out.

The AI shift is not a downturn. It's a structural change in what measurement looks like. Different kind of tide.

Over the next two to three years, three things are going to happen at the same time, and the dealers who can't close the loop are going to feel all of them at once.

One. AI lowers the cost of sophisticated attribution to a point where "we don't know where our customers come from" stops being a defensible answer. The agencies and dealer groups that have closed the loop will be able to point at exactly which dollar sold which car. The ones that haven't will be competing on gut.

Two. Ad platforms become more expensive and more opaque. Meta and Google are tightening their own attribution, optimizing for their own definitions of conversion, and showing less of the picture to the buyer. The dealers who own their data can push back. The ones who don't will be price-takers.

Three. Consolidation accelerates. Public groups — the Lithias, the Asburys, the AutoNations — have invested more heavily in data infrastructure than single-point operators, and the asymmetry is already visible in their cost-per-sold-unit. Lithia's Driveway platform³ reports unit counts across its originated deals as a first-class metric — not a guess, a count. That's what a closed loop looks like inside a large group. When a single-store operator competes against that, the asymmetry compounds every month.

The tide is going out right now. Not theoretically. In real stores, on real P&Ls, over the next twenty-four months. And when the tide is out, the dealers who've been running on vendor reports and gut feel are going to find out — publicly, and in the numbers — that they've been swimming naked.

VI. What the closed loop actually looks like

Let me describe this functionally instead of architecturally, because the architecture is a vendor conversation and the function is what matters.

A closed loop answers one question:

Which ad dollar, spent on which platform, through which campaign, against which audience, originated the customer who bought which car on which day — and at what gross?

It answers that question for every unit that changed hands last month. Every one. First-touch ad impression timestamped. CRM record linked. Test drive attributed. Deal logged. Gross attached. One row per sold unit. Every column filled.

Whatever the picture turns out to be, it will rarely match the gut. That's the value — not the surprise itself, but the fact that you were operating blind, and now you aren't. The reallocation that follows is usually more valuable than the attribution project itself. Vendor conversations change too: the ad agency, the CRM provider, the BDC — every one of them now has a single number they're accountable to, tied to the only outcome the business actually cares about. The conversations stop being about impressions, clicks, and "engagement," and start being about attributed units.

Three moves you can make Monday morning

You don't need to buy anything to start this work. Three moves, in order.

One. Reconstruct one unit manually. Pick a car you sold last week. Trace every touchpoint you can find — the first ad impression, the form fill, the call, the text, the test drive, the close. Use only the tools you already have. See how far back you can actually get. The gap between what you can reconstruct and what a closed loop would show you is the problem.

Two. Map your vendors and your data on one page. In a single table, list every vendor in your stack, what data lives in their system, and whether that data flows anywhere else. Most dealers have never done this on paper. Once you do, the handoffs that are missing become obvious in about fifteen minutes.

Three. Assign the seat. Pick a named person — not a role, a person — who owns the question which ad dollar sold which car? and who reports to you on it monthly. If you can't name that person today, you have your answer about why the picture doesn't exist.

What it costs, what it takes

This is work. The investment isn't trivial and the timeline isn't instant — a proper closed loop is a 60-to-120-day build, not a weekend project. The return is proportional to the discipline, not to the tech spend. Dealers who spend heavily on tools but never assign the seat end up with the same silo they started with, just with a more expensive login screen.

None of this requires you to become a data analyst. It requires you to decide that one question — which dollar sold which car? — is worth answering, and then either build or buy the infrastructure that answers it.

VII. The pushbacks I hear, and why they miss

When I argue this to dealer principals, I hear three versions of the same objection.

"I tried something like this and it didn't work." This one is the most common, and it's fair. Most attempts at closed-loop attribution in automotive have been sold by vendors with a narrow view and a narrow tool. A call-tracking company sells call-tracking attribution; it's incomplete by definition. A Facebook agency sells Facebook-reported attribution; it's self-serving by definition. A CRM vendor sells CRM-sourced attribution; it's blind to everything upstream of the lead record. If you tried one of these and it didn't work, you were right — it didn't work. But the category you tried was a point solution. The closed loop is not a point solution. It's a system.

How to grade anyone's closed-loop claim — including mine

Because this essay is written by someone who sells what it's arguing for, the honest move is to give you the test. Any vendor — me, or anyone else — who claims to close the loop should be able to answer yes to three questions. If they can't, the loop isn't closed.

One. Can you show me one row per sold unit last month, with the originating ad impression, the campaign, the channel, the lead record, and the gross — on one screen? Not in a dashboard that aggregates. In a list, deal by deal, that I can audit against my DMS.

Two. Do you own both sides of the loop, or just one? A vendor who handles only the ad side, or only the CRM side, is telling you half the story by definition. The loop closes end to end or it doesn't close.

Three. If I fire you, does the data leave with you? A real closed loop means the dealer owns the picture. If the vendor's answer is "we keep the data," the loop is closed to their benefit, not yours.

A vendor who passes all three is doing real work. A vendor who fails any one of them is selling you a fragment and calling it a system. That framework is worth more than the rest of this essay combined.

"I don't need it — I've been selling cars for thirty years." The past is the evidence for this argument, and the past is about to stop predicting the future. If the next thirty years of automotive look like the last thirty, this objection wins. They won't. AI-native groups are entering the market with structurally lower overhead and structurally better data discipline. You don't have to believe my ratios — you have to believe the direction. The thirty years of experience that got you here is real and valuable. It's not the thing that gets you through the next five.

"I don't understand it, and I don't want to." This one is the most honest, and it has the cleanest answer. You don't need to understand it. You need to decide that someone at your store owns the question which ad dollar sold which car? — and that person reports to you on it every month. Understanding the plumbing isn't the dealer principal's job. Owning the outcome is. Most dealers have never assigned the seat.

None of these objections are stupid. All of them have a version where they're right. The issue isn't that they're wrong in their moment. It's that the conditions they were right about are changing underneath them.

VIII. The next two years

I've spent the last decade watching the same mistake play out, month after month, in store after store. Dealers losing units they shouldn't have lost. Scaling ad spend that wasn't working. Killing spend that was. And — most of all — never getting a picture clear enough to tell which was which.

Dealer Ignition exists because I got tired of watching that happen and decided to build the marketing engine I wished my clients had. It runs the top of the loop: creative, media, campaign structure, the spend data that tells you which dollar went out the door and what it touched. Diablo AI exists because I got tired of watching the follow-up side rot for the same reasons — a dozen disconnected tools, none of them tied to the outcome that matters. It runs the bottom of the loop: lead capture, follow-up, conversation, conversion, the customer data that tells you which prospect came back through and which unit they bought.

Because both are built by the same team against the same data model, the loop actually closes end to end — ad impression to signed deal, one row per sold unit. Not as a feature. As the whole point.

You don't have to buy either of them. The test I gave you in Section VII applies to us exactly the same as it applies to anyone else. What you do have to do is decide that the question matters, and that somebody at your store owns it.

The next two years are going to decide which dealers can answer that question and which can't. The ones who can are going to spend their advertising with a precision that compounds every month. The ones who can't are going to spend more, hope harder, and watch the tide go out.

The loop is not going to close itself. Somebody at your store has to decide to close it. It might as well be you. It might as well be now.

---