Many are already starting to draw their own conclusion about the role of ChatGPT Ads for multifamily marketers. Our approach to this new platform has been different. We've decided to put them to the test and be open about the entire process.
We started by sharing 10 things we learned from setting up our first series of ChatGPT Ads for an apartment community. Now, as promised, we've got the first month of performance data.
Let us be clear up front: This is one apartment community with a $25/day budget for four weeks and $351 in total spend. It's a small test sample on a brand new ad platform, and still well within the optimization window of any new Google or Meta campaign.
But it's real money against real ad inventory. So, instead of making assumptions, we'd rather show what's actually happening—and not wait until there's a perfect answer to the question: "Should I run ChatGPT ads for my apartment communities?"
Here's how things are going:
ChatGPT showed the community's ads 3,531 times in a month with a CPM of $99.52.
There's no benchmark to hold that against yet. And because ChatGPT relies on context hints rather than keyword targeting, it's likely that the community's ads surfaced in answers that weren't necessarily direct apartment searches. So, we're just reporting the total number of impressions and cost per thousand impressions, not judging it.
The average CTR was 2.92%.
For context, the average CTR of this community's non-branded Google Ads and Bing Ads campaigns was around 4.45% at the end of the same period. ChatGPT Ads earned attention and showed marked improvement over time, a hallmark of the optimization phase that takes place within other established ad platforms.
103 people clicked on the ad. 
Clicks rose sharply, but then flattened towards the end of the month.
By the final week, visitors were spending about 10 seconds actually engaging with the website.
The first week total of just five clicks is a little misleading. The average engagement time from those clicks was 65 seconds, which was longer than the average engagement time from the community’s organic traffic.
Two of those five clicks qualified as Engaged Visits because those individuals completed a series of conversion actions on the community's website (think opening a floorplan page, watching a virtual tour, or clicking a form).
Were we onto something here?!
As clicks rose, however, their engagement quality weakened significantly.
In total, the first month resulted in 19 Engaged Visits—the most coming in the third week with seven. Out of 103 total clicks, that means 18 percent of ChatGPT ad traffic showed genuine engagement and could be considered more qualified.
Four floorplan views for the month. None in the final week.
Further corroborating those weakened engagement patterns is that there were only four floorplan page views the entire month, and none in the last week. That's an early challenge for ChatGPT Ads, especially in multifamily marketing, since opening a floorplan page is a website action that usually indicates higher leasing intent.
An engaged visit cost $18.49 on average.
Even though the ads didn't generate much traffic their first week, we were encouraged to see how quickly the new ad platform was able to produce engaged visits at a cost that was on par with this community's Google Ads campaigns, whose cost per engaged visit was $10.63.
But as quality diminished over time, costs increased.
Raising our bid ceiling didn't change much.
Average CPC remained stable at $3.41. That number was likely influenced by us initially setting the max CPC at $3.50. After the first couple of weeks, OpenAI suggested it be bumped up to $4, which we did. Impressions and clicks peaked in the third week, but then dropped off again. Even with the higher ceiling, our actual CPC never got near it, as it was $3.17 in the final week.
Alongside that, half of the roughly $700 available that month went unspent. There simply weren't enough matching conversations to deploy the community's ad against.
What does this all mean?
Again, one community, one month of ad spend. It isn't large enough of a sample size to draw any final conclusions from. This community's ChatGPT Ads are still active, so we'll continue following their progress (and soon others' as well).
But click quality, floorplan page views, and cost per engaged visit are all currently trending the wrong way. Part of the issue this early in ChatGPT Ads' existence is the limited targeting control outside of context hints and location.
That leaves the ad copy itself as the next lever to look at. So, for the next month, we’re going to update the context hints and rewrite some of the individual ads to give prospects clearer signals about the apartment community before they click. Stay tuned to see if engagement times and floorplan page views recover.
We are eager to see what enhancements OpenAI makes to its ad platform over time and how much more control advertisers will gain to adjust and optimize performance. For now, we'll keep the ads running. If we stop, we'll say so.
Interested in joining the test group?
If you're reading this and using our Predictive Advertising solution, we have everything in place for you to start running ChatGPT Ads, with in-platform AI visibility reporting coming soon. A larger sample size would benefit all of us.
Campaign management and analysis from Brandon Scholting, Digital Ads Team Manager, and Riley Northup, Senior SEO and Digital Ads Strategist. Questions about the data? Contact marketing@rentvision.com.