Third-party listing sites took $117,249 per dealership in 2025, 20.0% of all advertising spend, according to NADA’s annual report. Only search engine marketing, at 21.1%, took more. At that average, a group with eight rooftops spends close to $1 million a year on listings. The report you get from the listing site every month can’t tell you whether any of it worked.
The report counts what the site touched. It can’t count what the site caused. Those are different numbers, and your money goes into the gap between them. The test that answers “is Autotrader worth it” is cost per deduplicated sold match by store, set against what those shoppers would have done without the listing.
What does the listing report count?
Leads, calls, click-throughs, vehicle detail page (VDP) views, “connections.” Every one of those is a touch. None of them is a cause.
A shopper who found your used Tahoe on your own website, then saw it again on a listing site and submitted a lead there, shows up in the vendor’s report as a conversion. The site touched that shopper. It didn’t create them. Your inventory did, or your price did, or your store’s reputation did. The report can’t tell that shopper from one who had never heard of your store, and the vendor has little incentive to try.
Cox Automotive’s 2025 Car Buyer Journey study of 2,300 buyers found that 75% used third-party sites such as Edmunds, Cars.com, Autotrader, and Kelley Blue Book, including 70% of new-vehicle buyers and 77% of used buyers, and that buyers visited 4.6 websites of all kinds on average. Cox owns Autotrader and Kelley Blue Book, so read that as a vendor’s figure. Even at face value, it points to overlap. If three out of four buyers pass through a third-party site and the average buyer visits between four and five sites, many of your sold units were touched by several channels. Each channel’s report will claim those units. Add up the claims across your vendors and you’ve sold some of your inventory twice, on paper.
If a third of the sales a listing site claims would have happened anyway, your real cost per added sale runs 50% higher than the report implies. At $117,249 per rooftop, you need to know which third.
What are you paying, site by site and store by store?
Before you can judge the output, pin down the input. Pull every invoice: base package, premium placements, add-ons, lead products, trade tools. Per store, per month, per site. Then put that number on one page.
The published averages give you a rough check on your own invoices. Cars Commerce reported monthly average revenue per dealer of $2,500 across 19,343 dealer customers at June 30, 2026. CarGurus reported US quarterly average revenue of $8,134 per subscribing dealer in Q2 2026, across 26,151 paying US dealers. Neither is a rate card. The Cars Commerce figure is monthly and leaves out digital advertising services and DealerClub, and the CarGurus figure is quarterly. A store buying only the base package may sit below them, and one stacked with add-ons may sit well above.
Now build a simple table: site, store, monthly spend, annual spend. Add a column for contract end date and cancellation terms. You’ll use that column later.
What is a deduplicated sold match?
It’s the honest version of the vendor’s math, run by your side of the table:
- Export every lead and tracked call from each listing site for the trailing 12 months, by store.
- Match those contacts against sold records in your dealer management system (DMS), by name, phone, and email, with a reasonable window between contact and delivery.
- Check your customer relationship management (CRM) system for an earlier contact from any other source on every match. A website form two weeks before the listing-site lead means the listing site didn’t originate that buyer.
- Count each sold unit once, credited to the earliest contact, and divide each site’s spend by its deduplicated matches.
The number that comes out is cost per deduplicated sold match, by site and by store. Run it separately for new and used.
Expect friction. Business development centers (BDCs) log calls inconsistently, shoppers use a spouse’s phone and a work email, and leads that never reached the CRM can’t match at all. Matching is imperfect, so apply the same rules to every site, and the comparison between sites stays fair even if the absolute counts wobble. GA4 and your CRM won’t agree on these numbers either, so the DMS stays the record of what sold.
Expect the number to be uncomfortable. Say the vendor report claims 40 influenced units last month and the match credits the site with originating 12. Divide the invoice by 12 and you have a cost per sale you can set against what the next car earns and against every other channel you buy.
No independent study I could find measures shopper overlap between these listing sites. Nobody outside your DMS knows how many of your CarGurus buyers also submitted a lead on Cars.com, which is why the match has to run on your data.
What about shoppers who never submit a lead?
The match has a blind spot, and this one cuts in the vendor’s favor. A shopper who sees your truck on Autotrader, never submits a lead, and walks in Saturday morning leaves no trail. The listing earned that unit and gets no credit. Cut sites on the match alone and you risk cutting value you can’t see.
So test it. Pick one site and one market where the contract allows a pause. Pause the store’s listings for a set period, long enough to cover your typical shopping cycle, and compare sold units and gross against the store’s prior run rate and against your other stores. Hold everything else steady during the window: pricing rules, inventory mix, ad spend elsewhere. A multi-rooftop group has an advantage a single-point store never will because your other stores are the control.
If sold units hold while the listing is dark, the walk-in blind spot was smaller than the vendor claimed. If units sag, the site earns its invoice at that store, and now you know with evidence. Either answer pays for the test.
How do you decide: keep, cut, or renegotiate?
By store, with new and used apart. Group-wide averages hide the stores where a site earns nothing behind the one store where it earns plenty.
- Keep, where cost per deduplicated sold match beats your other channels and the pause test confirms added units.
- Cut, where the match shows a handful of originated sales against a five-figure annual spend and the pause test shows no sag.
- Renegotiate everywhere in between. Bring your deduplicated cost per sale and the published per-dealer averages to the renewal, and strip add-ons the match can’t defend.
A used-heavy store in a dense market may justify a premium package. A new-car store with strong original equipment manufacturer (OEM) traffic may need the base listing or nothing. The answer was never going to be yes or no for the whole group. It’s a per-store answer that changes with your inventory and your market, so run the match quarterly.
Who owns this work?
Not the rep selling the package. The rep’s income depends on your renewal, and every report the rep hands you supports it.
Not the BDC manager, whose scorecard is appointments and who has no authority over a six-figure contract. Not the general manager (GM) of one store, who sees one rooftop’s data and can’t run a test with controls. The work spans DMS exports, CRM hygiene, contract terms, and budget authority across every rooftop, and it has to answer to the principal.
A group of three to 15 rooftops usually has nobody whose job that is. The stores have marketing budgets. The group has marketing spend. Nobody owns the measurement, which is how a 20% line item runs for years on vendor-supplied evidence.
I’m a fractional chief marketing officer (CMO). I spent three and a half years running marketing for La Mesa RV, an 11-store RV dealer group, where I moved $500K a month out of broadcast that couldn’t be measured and into digital that could. That was RVs, not cars, but the method transfers: establish what each dollar can prove, cut what can’t prove anything, and build the measurement before the renewal date. A fractional CMO owns the deduplicated match, the pause test, the renewal negotiation, and the quarterly rerun, and reports the results to you, not to the vendor. If your listing-site line item has never survived that examination, it’s overdue.