ChatGPT just launched ads in Canada. Here is what real estate agents need to know.

ChatGPT launched advertising in Canada this month, and the way the targeting works is different from anything agents have used before.

On Google or Facebook, advertisers select from preset audience categories such as demographics, interests or behaviours. On ChatGPT, you describe the types of conversations you want your ad to appear alongside. Not who someone is. What they are currently thinking about.

Someone typing into ChatGPT asking what homes are selling for in their neighbourhood. Someone asking why their listing has not moved. Someone working through whether they can afford to upgrade before their kids change schools. The ad appears in the chat window alongside the AI’s response, inside a conversation that is already underway.

In a recent episode of their weekly sales and marketing show, Andrew Fogliato of Just Sell Homes and Real Estate Magazine broke down the platform mechanics and what the opportunity looks like for agents willing to test it early.

How the platform works right now

The setup is at ads.openai.com. Canadian agents can run ads targeting all of Canada – country-wide only, with no regional or city-level targeting available yet. The minimum spend is $25 a day.

To run ads, a business account needs to be verified, which requires a corporation number and HST/GST number. Verification takes several days. Once verified, the landing page the ad links to must be on the advertiser’s own domain – standalone pages on third-party tools will not be accepted. The advertiser’s website also needs to allow AI crawlers. Sites that block AI indexing, which some businesses have done to prevent their content from being used to train AI models, cannot run ChatGPT ads until that block is removed.

The ad itself appears below the AI’s response in the chat window. It includes a company name and a short text description. The amount of copy available is limited.

“The biggest question, assuming the context window works for getting in front of those people where we want, is user behaviour,” Fogliato said. “It’s different from every other ad platform. How do we get them to leave the chat and talk to you?”

The targeting logic

Where the platform differs from anything else available to agents is in how targeting works. Rather than selecting from a menu of audience attributes, the advertiser fills in a text field describing the conversations they want to appear in. ChatGPT then determines when a user’s session matches that description.

Fogliato walked through what that looks like in practice, using ChatGPT itself to identify the types of questions that signal intent.

Strong signals include conversations around home valuations, comparing agents, questions about the listing process and commission, questions about preparing a home for sale, capital gains on investment properties, timing the market and requests for comparative market analyses.

Medium signals include neighbourhood price trend searches, downsizing or upsizing content, retirement planning and relocation research.

The distinction matters because the intent behind a strong signal conversation is more immediately commercial. Someone asking “how much are homes selling for in Windermere right now” is likely a homeowner who is thinking about selling. Someone asking about retirement planning may be thinking about it years from now.

“Anyone asking about homes on Realtor.ca – because what they might do is take a listing from Realtor.ca, paste it in, and say: here’s my neighbour’s house. My house has this, this and this compared to theirs. What would my sale price be?” Fogliato said. “That’s a really interesting conversation to show up in.”

Two conversation types Fogliato flagged as particularly valuable for agents who work expired listings: people asking why their house is not selling and people asking how to get out of a listing agreement.

The upstream opportunity

The reason the platform is worth watching is not the targeting precision, which is limited at this stage, but the nature of the conversations happening inside it.

On Google, most search queries represent someone who has already moved through the research phase and is ready to act. The competition for those clicks is high and the cost reflects it. On ChatGPT, people are often working through earlier stages of a decision. They are describing a situation rather than searching for a solution. They are asking questions they would not type into a search bar.

Taylor Hack described the dynamic using the example of a family six months after their second child’s birth when both incomes are back, the immediate costs of a new baby have settled, and the mismatch between the house they have and the house they need is becoming clear. That person might not be searching for a real estate agent. They might be asking ChatGPT questions about school divisions, costs, or what their current home is worth.

“This is showing up in upstream conversations,” Hack said. “Before they’re in action mode. Before they’re searching for an agent.”

Fogliato also pointed to a connection between paid placement and organic visibility. On Google, running paid ads to a page has historically coincided with faster organic ranking improvements for that page, even though Google has maintained the two systems operate independently. Whether a similar dynamic will emerge on ChatGPT is not yet known, but it is a consideration for agents who are also building content intended to be cited by AI search.

What the call to action problem looks like

The open question at this stage is whether users will consistently leave a ChatGPT conversation to click through to an agent’s website.

Fogliato drew a comparison to the early days of Instagram advertising, when users were not accustomed to leaving the app to visit external links. Traffic ads performed poorly until the platform introduced the swipe-up feature and a built-in browser. ChatGPT ad users may behave similarly, particularly for searches that feel informational rather than transactional.

His hypothesis: a call to action that offers something specific and immediately useful – a list of homes, a neighbourhood price guide, a single question answered – is more likely to generate a click than a general invitation to get in touch.

“Will they want to go to an informational thing? Will they only leave to book a time to talk to you?” he said. “Are they getting everything they need out of chat and they don’t want to leave until they’re ready to have that conversation?”

Who should be testing this now

Fogliato was straightforward about who this is and is not for at this stage.

Agents with the budget, the capacity to manage incoming leads and the willingness to test something with no benchmarks yet are in the best position to move early. The $25 daily minimum is a manageable test budget for most. The setup requires some technical steps such verifying a business account, ensuring the site allows AI crawlers, building a landing page on the agent’s own domain, that not every agent will want to navigate immediately.

“If you’re well-positioned for this opportunity, if you have ad budget to spend, if you can work your way through this kind of thing, and if you have capacity to manage the outcome and measure it – you’re an early bird,” Hack said. “If you don’t, wait until other people show you the way. Don’t spend test money.”

Fogliato and Hack set up Taylor’s first ChatGPT ad account live during the episode and are planning to share results in a future episode.

Join the conversation live

Each week, The Leads Are Sh*t brings working agents together to dissect what’s actually working in real estate right now.

Be part of the live recording, get your questions answered and join the after-show conversation.

👉 Register for the next live session.

Related Posts