The ChatGPT Ads Test That Rewrote a Toyota Dealer's Media Mix

Every dealer already spends on Google, Meta, YouTube, and CTV - the same channels every competing store in the market is running too. So when a single-point Toyota rooftop in the San Francisco Bay Area started running ChatGPT Ads, the question wasn't whether the channel could deliver clicks. It was whether those clicks would turn into anything that looked like real shopping behavior, on a budget too small to waste finding out the hard way.

The real problem

ChatGPT Ads was a brand-new advertising channel with no track record for this dealer, or for the auto vertical generally. Every dollar of the fixed $500 monthly budget needed to answer two questions before the dealer would commit more spend behind it: would real shoppers click, and once they landed on the site, would they act like shoppers - browsing inventory - or just bounce? On a budget this size, every wasted click was learning, and money, the dealer wouldn't get back.

What we built instead of one generic ad set

Clover ran a 63-ad program across five intent-based ad groups - Used & CPO Inventory, New Vehicle Interest, Geo/Conquest, Brand/Value Prop, and Financing - instead of one ad set trying to cover every shopper. The program ran under daily management for the full 31-day flight, August 1–31, 2026:

• Bought on Clicks, not Reach. The flight ran on a Clicks (CPC) objective, so the dealer paid only when someone clicked - every result below is a read on real shopper behavior, not just how many people saw an ad.

• 8 bid-cap adjustments across 6 dates held delivery in the band between collapse and overspend, after an early over-aggressive cut briefly stalled things.

• Every landing page instrumented past the click - vehicle search results and vehicle detail pages, not just impressions and clicks — so creative calls could be checked against what shoppers actually did next, not click-through rate alone.

What a full month actually delivered

$490.99 of the $500 monthly budget went out the door - 98.2% delivered, on 14,896 impressions and 201 clicks. That's a 1.35% click-through rate at $2.44 per click.

Those 201 clicks generated 1,318 Interactions - tracked page views on a click-attribution window, about seven pages per visit on average. Interactions aren't leads: form submissions, phone calls, and appointments aren't tracked on this account yet, and we're not going to pretend otherwise.

Here's the number that matters most: 42.4% of every Interaction was a shopper looking at inventory - a vehicle search results page or a vehicle detail page, not a homepage bounce. That's the strongest inventory-intent read of any store this dealer runs.

Metrics & Results

• Budget delivered: $490.99 of $500 (98.2%)

• Impressions / clicks: 14,896 / 201

• Click-through rate: 1.35% at $2.44 avg. cost per click

• Interactions tracked: 1,318 (about 7 pages per visit)

• Share of Interactions that were inventory research: 42.4%

Two ads alone drove 41% of all clicks and 52.5% of all Interactions - a used-inventory search ad and a price-filtered used-inventory ad, on well under half the month's budget. The account closed the month demand-limited, not budget-limited: cost per click stayed well under the bid cap through the final days, so September's plan is to add more of what already works rather than spend more to chase the same result.

Client perspective

"ChatGPT Ads is the first new channel we've tested where the users who clicked kept browsing our inventory instead of just bouncing. This is the kind of proof we needed before we put more budget behind it to double-down on what's working. I see this being one of many channels that will play an integral part in our media mix moving forward." -Toyota Marketing Director

What this means if you're deciding whether ChatGPT Ads earns a spot in your mix

A full month of real data answers the question week one couldn't: this wasn't a novelty click. Nearly half of every tracked interaction on this account was a shopper actively working through inventory - the exact behavior a dealer needs to see before committing more budget to a brand-new channel. And the fact that two ads carried over half the results on well under half the spend says the next dollar has an obvious place to go.

The differentiator isn't the platform. It's running it like a real media buy from day one - intent-based ad groups instead of one flat set, daily bid management instead of set-and-forget, and a read on performance that goes past the click. That's true whether you're running ChatGPT Ads specifically or evaluating any new channel where "we tried it once" is the difference between a real read and a wasted budget.

Curious what running ChatGPT Ads would look like for your store? Schedule a demo and we'll walk through what we'd build for your market.