// sector · Real estate
Before calling an agent, a buyer tells ChatGPT about the project: the city, the budget, the number of bedrooms, the timing. Your offer can be in that conversation, and we measure what it triggers, down to the appointment.
we're looking for a 3-bedroom flat near Lyon, budget around €450,000, new build or older property?
At equal budget, the question is less new versus old than how long you can wait and how much you're willing to redo. A new build is delivered later but needs no work; an older flat can be visited now and needs a renovation budget. Before viewing, fix your move-in date, the amount of work you'll tolerate and how far from the centre you'll go.
3-bed new builds, Lyon
Brochure, prices and floor plans. Delivery date.
your-development.fr
// why this sector
A property project fits in one sentence: where, how much, what kind, by when. Nobody types that into Google. Everybody tells it to ChatGPT.
A job transfer, a separation, a first purchase as a couple: the project is born from an event, and the buyer tells it with the area, the budget, the number of rooms and the timing. A query like "flat Lyon" contains none of that. A prompt holds the entire brief, and a context hint can describe it.
"We're just looking" and "we sign the offer next week" don't call for the same ad. Hence ad groups split by maturity (discovery, comparison, valuation, appointment), because a buyer who is comparing doesn't read the ad meant for a buyer ready to sign.
A hint can say "family looking for a three-bedroom flat near Lyon", but it cannot say "only show in Lyon". The country is set at campaign level; the city describes the buyer without filtering delivery.
// intents
Typical prompt
"First purchase, tight budget, does a new-build scheme make sense for us?"
What the ad has to do
Name the situation (first-time buyer, constrained budget) and lead to a page that compares honestly, rather than to a list of units.
Typical prompt
"Sell my house myself or go through an agent, what actually changes?"
What the ad has to do
Carry a single verifiable argument per ad. The natural conversion is the valuation request.
Typical prompt
"Buying a small flat to rent out, which mid-sized city and what budget?"
What the ad has to do
Describe an investor's situation with no promise of returns. The landing page shows real properties and offers a simulation on request.
Typical prompt
"We're relocating for work, what is our flat worth and how long does it take to sell?"
What the ad has to do
Answer the immediate need, a valuation and an appointment. The ad leads to the form, not to the homepage.
// context hints
A property project is told through a place, a budget and a moment in life: the move for work, the separation, the first purchase. The hints keep that thread without pretending to filter by city, grouped by ad group, as in this extract for a developer:
Generate your context hintsIllustrative excerpt. The real structure is built from your search terms, sales calls and support tickets.
AG · Buying a home
AG · Buy-to-let
AG · Selling and valuation
// scenarios
Three examples that tie a conversation to an ad, a page and a metric. They are not delivery promises.
// measurement
In real estate, the decision is made on the appointment that follows the click and the mandate that comes out of it, two steps OpenAI's native report cannot see. We wire the OpenAI Pixel and the Conversions API to those moments, then place the account against the Hiints Index.
Valuation requests
The first signal for a network. We separate the serious valuation from the curious one right at the form.
Appointments booked
Viewing, sales office or advisory call: the attended appointment is the conversion that counts, not the form.
Brochure downloads
Useful for new builds, provided you track in the CRM what becomes of those contacts.
Qualification rate
The share of contacts whose budget and timing match the offer. An off-budget lead costs more than a lost click.
Cost per mandate or reservation
The number the owner cares about, to be compared with your portals and Google Ads over the same period.
Your costs vs property accounts
The CPC of a new-build scheme and that of a valuation request can't be compared with each other. We place each against comparable accounts in the Hiints Index.
// honest qualification
Yes, clearly
You sell new-build schemes, run an agency network or a property-management offer, and a page per offer exists or can exist. Your conversions are clear (valuation, appointment, brochure), and you run nationwide, or you accept that the city is a context, not a filter.
Yes, with one caveat
An independent agency in a single town runs into country-level targeting: some clicks will come from elsewhere. We then look with you at whether a remote valuation offer or a regional scope makes it worthwhile. If your offer touches on credit or investment products, it falls under OpenAI's advertising policies and its eligibility is checked before any launch.
Not yet
With no way to follow a contact after the form, nobody will ever know what got signed, and the test would teach nothing. The finding is the same under €3,000 of monthly media, for lack of volume. In both cases, you're better off staying on your portals.
Free setup, done in under 48h. Then 15% of ad spend under €5,000 of monthly media, 10% above. See pricing →
// questions
Not as a delivery rule: the campaign targets a country. The context hint then describes a situation, for instance "family looking for a three-bedroom flat near Lyon", and steers matching towards conversations that resemble it without restricting delivery.
The brochure download is only a start. We then tie clicks to sales-office appointments and reservations in your CRM, through the OpenAI Pixel and the Conversions API.
One ad group per scheme or per project family, with its own hints and a dedicated landing page. The ad names the buyer's situation, and its title fits in 24 characters.
No. "3-bed flat Lyon" is a query, not a situation. We rewrite each intent as a natural sentence, the way a buyer phrases it inside ChatGPT.
We don't answer on OpenAI's behalf. Anything touching on credit, investment products or tax schemes falls under its advertising policies, and the eligibility of the offer and of the messages is checked before any launch. The ads never promise a return.
Everything starts from the value of a mandate or a reservation: it tells you how many off-target contacts you can absorb before judging the channel. A mandate is signed after the appointment, so the test must run long enough to wait for it. Setup is free, and the fee follows your media (15% under €5,000 a month, 10% above). Below €3,000 of monthly media, we don't recommend testing.
// keep reading
Context Hints in ChatGPT Ads: Definition, Format and Examples
What a context hint is according to OpenAI, where it is set, the who/when/what format, five good-versus-bad examples and the mistakes to avoid.
Read →How-toHow to Structure a ChatGPT Ads Account: Campaigns to Ads
One campaign per objective, one ad group per intent family, one ad per angle. Naming rules, intent families and a complete example for a B2B SaaS.
Read →ConversionChatGPT Ads Landing Pages: The Page That Continues the Conversation
The landing page is one of OpenAI’s ad selection signals. One page per family of intent, the checklist, three prompt → page pairs and message match.
Read →// shall we talk?
Thirty minutes on a call is enough to set your segment against the Hiints Index, live. If ChatGPT Ads doesn't serve your business, we'll say so before you have spent a euro.
Book a call30 minutes · no commitment · zero pitch
// other sectors
