More clicks are not more customers. One Genesis dealership in New York State proved it for us this year: after moving its Google Ads to an OEM-certified digital vendor, its paid traffic grew 77%, and its calls and forms stayed flat.
Every dealer eventually faces the same choice. Go with the vendor your OEM certifies, and collect the co-op dollars that come with it, or stay with an independent agency. Both can be the right call. But the decision deserves real numbers, not a feeling, so here are ours.
The setup
SmartSites ran this store’s Google Ads, both paid search and Vehicle Listing Ads (VLAs), until the dealership moved to the OEM-certified vendor. The program came with the certification, and the store made a reasonable business decision.
The website and its Google Analytics 4 (GA4) property stayed the same, which gave us a clean before-and-after. We compared the same eight months, March through October, in 2025 (SmartSites) and 2026 (the certified vendor). That removes seasonality, which matters in car sales.
We counted only Google Ads traffic. A lead here means a click-to-call from the website, or a sales, appointment, service or chat form submission. These are the same standard dealer events, tracked the same way in both years.
Same leads, 77% more traffic
In 2026 the store got 293 calls and forms from Google Ads, against 286 in 2025. Getting there took 49,864 paid sessions instead of 28,205.
| All paid Google traffic | 2025 (SmartSites) | 2026 (certified vendor) | Change |
|---|---|---|---|
| Sessions | 28,205 | 49,864 | +77% |
| Engagement rate | 71.0% | 61.4% | −9.6 pts |
| Avg. engagement time per session | 69s | 38s | −45% |
| Pages per session | 2.55 | 1.92 | −25% |
| Calls | 197 | 171 | −13% |
| Forms | 89 | 122 | +37% |
| Calls + forms | 286 | 293 | +2% |
| Calls + forms per 1,000 sessions | 10.1 | 5.9 | −42% |
The telling number is total engaged time: 32,323 minutes in 2025 against 31,140 in 2026. Shoppers spent about the same total time with the store’s inventory. The extra 21,000 sessions added almost nothing, because those visitors left quickly.
Forms did rise in 2026, and that deserves credit. But calls fell, and for most dealers a phone call is the lead most likely to turn into an appointment.
VLAs: more than twice the traffic, a third of the calls
The gap is widest in Vehicle Listing Ads. VLA sessions more than doubled in 2026, while VLA calls fell from 67 to 21.
| VLA / Performance Max | 2025 (SmartSites) | 2026 (certified vendor) | Change |
|---|---|---|---|
| Sessions | 13,427 | 28,635 | +113% |
| Engagement rate | 69.2% | 55.1% | −14.1 pts |
| Avg. engagement time per session | 51s | 24s | −53% |
| Calls | 67 | 21 | −69% |
| Forms | 46 | 47 | flat |
| Calls + forms per 1,000 sessions | 8.4 | 2.4 | −72% |
In June 2026, VLA engagement rate fell to 40%, meaning six in ten VLA visitors left without engaging at all.
A VLA shows a specific car, with its photo and price, to someone already searching for it. Done well, it is the highest-intent placement a dealer can buy. When VLA traffic doubles and leads don’t move, the campaign is usually reaching wider and cheaper: broader audiences, looser product targeting, or a goal of clicks and vehicle page views rather than calls and forms. We can’t see the vendor’s account, so we can’t say which. The visitor behavior shows the result either way.
Paid search: where the new program held its own
Paid search is closer to a draw, and in total leads the certified vendor came out ahead. It ran 44% more search traffic and turned it into 225 calls and forms, against our 173.
| Paid search | 2025 (SmartSites) | 2026 (certified vendor) | Change |
|---|---|---|---|
| Sessions | 14,778 | 21,229 | +44% |
| Engagement rate | 72.7% | 69.8% | −2.9 pts |
| Avg. engagement time per session | 85s | 56s | −34% |
| Calls + forms | 173 | 225 | +30% |
| Calls + forms per 1,000 sessions | 11.7 | 10.6 | −9% |
Conversion rates were close, and visitors stayed longer in 2025. Search works on what shoppers type, so a bigger budget there tends to buy more of the same intent. VLAs are different: the platform decides more of who sees them, which makes them easier to scale into low-quality clicks.
The calls GA4 never sees
Everything above comes from the website, and that leaves out a big share of the phone calls. Many shoppers tap the call button right on the ad and never visit the site, so GA4 never records them. Google Ads does, as “Calls from ads.”
From March to October 2025, our campaigns for this store drove 718 calls straight from the ads, 515 of them from VLAs. That is more than three times the 197 website calls above.

We don’t have the 2026 figure, because it lives in the vendor’s account. That is the point for any dealer reviewing a vendor: ask for calls from ads by campaign type, every month. If a report leaves them out, you are missing a large share of your leads.
Three measurement traps we hit, including one in our favor
A fair comparison took more cleanup than we expected. Watch for these whenever you compare one vendor’s results with another’s.
- GA4 can label the same campaign differently. In 2025, GA4 put our VLA campaigns under “Cross-network.” In 2026, it put the vendor’s VLA campaigns under “Paid Search,” most likely because their ad account isn’t linked to GA4 the same way. Read only the channel report and it looks like VLA traffic vanished, which it didn’t. We grouped campaigns by type, by name, in both years.
- New tracking makes a year look better than it is. Late in 2026, new call and form events were added to the site. More events means more “conversions” without one more customer. We counted only the standard events that were tracked the same way in both years.
- A spike that is too good to be true usually is. In March and April 2025, GA4 recorded 705 VLA click-to-call events on about 3,700 visits, about one call for every five visits. Google Ads’ own count of calls from those same ads stayed normal, at 29 and 37. That pattern points to a tag firing more than once or to bot clicks, not to real calls. Leaving it in would have more than quadrupled our call count. We took it out.
What to ask any vendor, certified or not
OEM certification tells you a vendor meets the program’s requirements. It doesn’t tell you how your ads will perform. Before you switch, or at your next review, ask:
- What are the campaigns optimized for? Calls and forms, or clicks and vehicle page views? The answer shapes every bid.
- What are leads per 1,000 sessions, by campaign type? Total traffic is the easiest number to grow and the least useful one.
- How many calls came straight from the ads? Ask for this monthly, split by VLA and search, from the Google Ads account itself.
- What happened to engagement rate and time on site? A falling engagement rate alongside rising traffic is the clearest sign you are buying cheaper clicks.
- Did tracking change between the two periods? New events, or a newly linked or unlinked account, can move the numbers without a single new customer.
- Do I own the Google Ads account and its history? If you leave, you should keep your data. Without it, you can’t run a comparison like this one.
The bottom line
This is one store over eight months, and the website events we counted are not CRM-verified sales. It isn’t proof that certified programs underperform, and we don’t believe that’s true in every case. It does show, clearly, that traffic is the wrong scoreboard: the same leads at nearly half the conversion rate mean the dealer is paying for a lot of clicks that never become customers.
Co-op dollars are real money. So is the gap between 8.4 and 2.4 VLA leads per 1,000 visits. Weigh both.
Want to know how your own Google Ads and VLA campaigns stack up? SmartSites will run the same year-over-year review on your account: leads per 1,000 sessions, calls from ads, engagement and tracking health. Get in touch and we’ll show you what we find, good or bad.
“We've seen substantial growth in organic traffic, non-branded search visibility, and overall digital engagement since working with them.”
Danielle ColucciToyota of Stamford




