A dealer group with eight rooftops gets eight Google Analytics 4 (GA4) reports. Each one was built by the vendor that runs that store’s website. Each one defines a lead differently, names its events differently, and grades the work of the company that produced it. Stack them side by side and you still cannot answer the only question that matters at the group level: did the money sell cars?
This is where the money is left on the table. Not in the ad spend itself, but in the architecture around it. The group pays for twelve measurement systems and owns none of them. The fix is one GA4 property, in the group’s own Google account, receiving the same standardized events from every store — alongside whatever each vendor runs for its own purposes.
I spent 3.5 years running marketing for La Mesa RV, an 11-store RV dealer group, where we moved $500K a month from unmeasurable broadcast into measurable digital. I have never worked at an auto dealer, and I don’t claim otherwise. But the problem — multiple rooftops, vendor-owned analytics, no group-level view — is the same problem, and so is the architecture that solves it. What follows is that architecture.
Why can’t I just read the vendors’ reports?
Four reasons, and each one compounds the others.
First, the event names don’t match. One platform calls a form submission one thing, another calls it something else, and a “lead” on Store 3’s report may include chat starts that Store 7’s report excludes. You cannot compare numbers that were never defined the same way. (Why GA4, the CRM, and the vendor reports never agree is the single-store version of this problem.)
Second, the properties belong to the vendors. Dealer Inspire, for example, runs its own GA4 property built to the Automotive Standards Council (ASC) specification — though it states dealers may maintain their own separate property as well (Cars Commerce help documentation). When the property lives in the vendor’s account, the history lives there too. Change website providers and the data does not come with you. Three years of behavioral history, gone at contract end.
Third, every report is graded by the vendor that produced it. The website provider reports on the website. The SEM agency reports on the SEM. Nobody in that arrangement has an incentive to show you cost per sold unit across the whole group, because that number might not flatter them.
Fourth — and this is the quiet one — there is no group-level denominator. Twelve reports with twelve definitions cannot roll up. The dealer principal is left comparing a feeling about Store 2 against a hunch about Store 9.
The good news: the industry has already done the hard standardization work. The ASC specification names events the same way across platforms — a form submission is the same event regardless of which certified platform fires it — and that shared naming is exactly what makes cross-store comparison workable (Marchex). Dealer.com sites are ASC-certified (Cox Automotive), and other major platforms have followed the standard. The pieces exist. What’s missing is a property the group owns to receive them.
What should the group actually own?
One GA4 property. In the group’s own Google account — not the vendor’s, not the agency’s. Every store’s website sends the same ASC-named events into it, in parallel with whatever property the vendor maintains for its own reporting.
Ownership here is not a metaphor. It means three concrete things:
- The group’s login. The dealer principal or marketing director is the account owner. Vendors get access granted to them, revocable, not the other way around.
- The group’s data retention. You set the retention window. The history survives a vendor change, an agency change, an OEM program change.
- The group’s export rights. The data can leave. Into a warehouse, into a spreadsheet, into whatever the next generation of analysis looks like. Dealership analytics ownership is the difference between renting your numbers and holding them.
One honest caveat: tag and script permissions differ by platform and by OEM program. Some certified platforms allow a dealer-placed GA4 tag freely; others restrict what runs on the page. Don’t assume — ask each provider, in writing, what its platform and your OEM program allow. Compliance with an OEM program is a conversation with that program, not something to treat as settled from a blog post. (What the OEM’s co-op rules do and don’t control is the wider map of that territory.) Dealer Inspire’s own documentation acknowledging a dealer-maintained separate property is a useful precedent to bring to that conversation.
What do we measure at the group level?
Sessions and VDP (Vehicle Detail Page) views per store are activity. Worth watching, not worth deciding on. A store can post record VDP views and sell fewer cars than last quarter.
The chain that matters runs past the website entirely: leads logged in the CRM (Customer Relationship Management system), appointments set, shows, and sold units from the DMS (Dealer Management System) — joined back to the source that produced each lead. GA4 is the behavior layer. The sale lives in the DMS. A group that measures only the behavior layer is measuring the top of a funnel and guessing at the bottom. (From click to sold VIN walks the four layers in detail.)
How do the two get joined? Two workable patterns, neither requiring a specific vendor:
- Carry a lead ID through. The web form generates an identifier that lands in the CRM with the lead and travels to the sold record. When the unit sells, the ID traces back to the session and the source.
- Matchback on the sold record. Monthly, take the DMS sold file and match it against the CRM lead file on customer identity — name, email, phone. Less precise than a carried ID, but it works with the systems most groups already have, and it answers the question at the grain that matters: units, by store, by source, by month.
Either way, the output is the number no single vendor will hand you: cost per sold unit, by store, by channel.
How do I compare my stores without being unfair to any of them?
Honest comparison needs four disciplines, and the fourth is the one most groups skip.
Same event definitions — which the ASC naming standard gives you. Same date ranges, always; a 28-day window at one store against a calendar month at another is noise dressed as insight. Cost per sold unit as the headline metric, by store and by channel, because it is the only number that normalizes a high-traffic metro store against a rural rooftop.
And fourth: a written note, per store, on what its OEM program restricts. This matters more than it sounds. A store that looks worse on paper may simply sit on a stricter platform — fewer tags allowed, a narrower event set, a mandated provider with less flexibility. If that context isn’t written down next to the number, the group will punish a store for its platform instead of its performance. One page per store. What the platform allows, what it doesn’t, dated, revisited when programs change.
What does this actually change for me?
Three things, in rough order of dollar impact.
Budget moves between stores on evidence. When cost per sold unit at Store 4 runs double the group median on the same channel, the money has somewhere better to be — and now you can see it. Across 3–15 rooftops, the reallocation opportunity is usually larger than any single vendor optimization.
Every vendor gets graded on one scorecard. Not their scorecard — yours. Website provider, SEM agency, third-party lead sources, all measured in the same group-owned setup, all rolled to the same sold-unit outcome. (The vendor scorecard itself is its own article.) Vendor conversations change permanently once the group holds its own numbers.
And the dealer principal gets, for the first time, a group-level answer to “did the money sell cars?” — not eight partial answers from eight interested parties. That answer is also the foundation for whatever comes next: AI search is emerging, attribution will keep shifting, and no one can guarantee how any of it shakes out. A group that owns its own measurement layer can adapt. A group reading twelve vendor reports cannot.
Where do I start? The first 30 days.
This is a 30-day project, not a replatforming.
Week 1: Create the property. One group-owned GA4 property in a Google account the group controls. Name the account owner. This costs nothing and takes an afternoon.
Weeks 1–2: Ask every provider two questions, in writing. Which ASC events does your platform send, and which tags or scripts does your platform and our OEM program allow us to place? The answers will differ by platform and program. Get them documented before anyone touches a tag.
Weeks 2–4: Get the sold data flowing. A monthly CRM lead export and a monthly DMS sold export, delivered on a schedule, to one place. Start with the matchback pattern — it works with files you already have.
Week 4: Build the one-page table. One row per store: sessions, VDP views, leads, appointments, shows, sold units, spend, cost per sold unit, and the platform-restriction note. One page. If it doesn’t fit on one page, it won’t get read in the Monday meeting.
Thirty days in, the group won’t have perfect attribution. It will have something more valuable: a single, owned, comparable view of the group’s analytics data that gets sharper every month and belongs to no vendor.
Who owns the measurement layer?
The work here is not technical. Creating a GA4 property takes an afternoon. The hard part is the decisions: what the group measures, getting answers in writing from every platform, holding vendors to one scorecard, and keeping the sold-unit join running month after month. That is exactly what a fractional CMO at a dealer group should own on the group’s behalf — building the measurement layer with the stores, answerable to the group rather than to any vendor’s interest, and making sure that when the next vendor change comes, the numbers stay with the people who paid for them.