Buyers now ask an AI which estate agent to call. Make sure it says yours.
Buyers in your market now ask ChatGPT, Google, Gemini and Claude who to trust with a purchase, in their own language, and the assistant answers with three or four agencies. If yours is not one of them, that enquiry goes to a competitor and you never hear about it. We find out what the assistants say about you today, then do the work that changes the answer, with your approval on every piece.
The audit is a snapshot: real assistant answers, sampled the moment you run it. The assistants change their answers from one week to the next, which is why the programme measures every week. Treat the score as your starting point.
The first question now goes to an AI.
Buyers used to start with Google and a portal. More and more of them now start by asking an AI assistant, and the assistant replies with three or four agency names. If yours is one of them, you get the call. If it is not, the buyer may never find out you exist.
of consumers used AI recommendations for local services in 2025, up from 6% the year before.
BrightLocal, Local Consumer Review Survey 2025of UK adults used a generative AI tool in the past year. Your British buyers and vendors are in that number.
Ofcom, Online Nation 2025homes were bought by foreign buyers in Spain in 2025. Many of those searches started abroad, in another language.
Registradores de España, 2025growth in American buyers in Spain in a year, the fastest of any nationality. Four in ten of them buy in Andalucía.
Registradores de España, 2025Which estate agents should I use to buy a villa in Marbella?
Agencies that come up often for villas in Marbella:
- 1Altavista Homes
- 2Costa Vista
- 3Azul Estates
Illustrative: the agency names are invented.
Three names for one question.
The acronyms keep changing. The question behind them does not: when a buyer asks an AI assistant who to call, is your agency in the answer? Everything on this page is about making that answer yes, and about knowing every week whether it still is.
The work that gets your agency read, trusted and named by the engines behind ChatGPT, Google AI Mode, Gemini and Claude. Most of it happens on the sources those engines read, and on the pages of your own site they can quote.
Writing your pages so an assistant can lift a direct answer from them: the question a buyer asks, a clear answer, the figures behind it, marked up so a machine can tell which is which. The on-site half of GEO.
How often the assistants name your agency when buyers in your market ask, which sources they trust when they do, and whether your own website gives them anything to read. One number a week, per assistant, per language.
Ranking is no longer enough
A Google results page has ten links and a buyer can scroll. An AI assistant gives three or four names and stops. Page one used to earn you clicks; outside the handful there is no position at all.
Buyers ask questions, not keywords
Nobody types "estate agent Marbella" into ChatGPT. They ask for an agent who knows family homes near international schools, in Dutch or in Swedish. Longer, more specific questions, answered differently in every language.
Trust is borrowed from the sources the engines read
The assistants mostly repeat what the pages they trust already say: portals, press, directories, reviews, your Google Business Profile. Your website decides whether you can be read; those sources decide which names come up.
We treat GEO and AEO as one job with one weekly number. The free audit shows where you stand today; the programme does the work.
Your AI visibility in three numbers, and exactly what to do about it.
These are the pages of the report we build for every agency on the programme, refreshed every week. The free audit gives you the first of them today: the three numbers from fifteen buyer questions and searches in up to two languages, the readiness checks and your first moves.
At a glance
Share of voice. Of every agency mention across 248 answers, 13 were you.
Named in answers. 13 of 248 answers named you; three of the four assistants did so at least once.
Readiness / 100. Whether the assistants can read your site at all.
Your site can be read (81 out of 100), yet when an assistant named an agency, yours was the name 4% of the time. Being readable gets you considered. It does not get you named.
See exactly who is winning the AI answers in your market
Every competitor, counted
Every answer is read for the agencies it names, so you see the real shortlist the assistants have settled on, not a guess.
How far ahead they are
Not a ranking position but a count: how many times each agency was named, and how many mentions separate you from the leaders.
The gap you can close
The names above you show who the assistants trust today, and point at the sources you need a presence on to join them.
Who the assistants name, across 248 answers
Each answer counts once per agency it names, with every spelling of the same firm merged first.
Each assistant has its own shortlist. Find out which ones you are missing from
Across a row: who names you
Which of the four assistants name your agency, and which never do. An empty cell is a group of buyers you are invisible to.
Down a column: who it prefers
The agencies each assistant keeps returning to for your market, and how often. Those are the names your buyers are given.
Where the work starts
The empty cells in your own row come first. Each one has a cause, and the rest of the report names it.
Who each assistant names, and how often
| ChatGPT | Google AI Mode | Gemini | Claude | Total | |
|---|---|---|---|---|---|
| Your agency | 4 | 3 | 0 | 6 | 13 |
| Altavista Homes | 20 | 13 | 10 | 9 | 52 |
| Costa Vista | 20 | 8 | 9 | 10 | 47 |
| Azul Estates | 18 | 6 | 8 | 7 | 39 |
| Casa Meridian | 11 | 4 | 4 | 6 | 25 |
| Solana Group | 10 | 7 | 5 | 2 | 24 |
ChatGPT is read through two routes, the API and the public app, so its column counts two answers per question.
Named you, by assistant
The half of the problem you control completely
21 checks, scored out of 100
Crawler access, structured data, local presence, authority signals and content depth: everything that decides whether an assistant can find, read and trust your site.
Every failed check comes with its fix
Named and specific: unanswered reviews, missing FAQ markup, a Google profile that has gone quiet. Ready for your developer, or done by us where you give us access.
The quickest wins on the page
Most readiness fixes take days, not months, and they count from the next time the assistants read your site.
Whether the assistants can read you at all
Checked from outside, the way an assistant sees your site.
The pages the assistants trust in your market, and whether you are on them
Every source the assistants used
Portals, local press, directories, forums and guides: the pages the assistants leaned on while answering your buyers' questions, ranked by authority.
Cited for a rival, never for you
The sources that only come up when a competitor is named are the ones you are missing from. That list is your placement plan.
Worth more the higher it sits
Authority scores how many other sites link to a source, from 0 to 100. A listing or a mention on a high-authority page carries much further than one on a weak page.
The sources behind the answers
sources the assistants cited while answering the buyer questions.
were cited in an answer that named you.
were only ever cited in answers that named a competitor.
| Source | Authority | Cited in | Names you | Names a rival | Kind |
|---|---|---|---|---|---|
| coastpropertynews.com | 34 | yes | yes | press | |
| buyersguide-spain.com | 29 | no | yes | other | |
| homes-directory.es | 40 | yes | yes | directory | |
| andalusia-forum.net | 22 | no | yes | community | |
| relocationreview.co.uk | 19 | no | yes | press | |
| regionalmagazine.es | 11 | yes | yes | press |
A plan in plain language, ordered by what closes the biggest gap first
Every move traces back to a page of the report
No generic checklist. Each move comes from a gap the measurement found: a question where a rival is named, a failed check, a source you are missing from.
Owners and dates
On the programme the plan becomes the first month's work queue, with what we do and what your developer does, and when.
Reviewed with you before anything is published
Nothing in the plan promises a mention; each move gives one its best chance. You see it, and approve it, before any of it goes live.
The first three moves
Answer the 36 buyer questions where a competitor is named and you are not
A page to publish or deepen for each, in the buyer's own language, briefed from the pages the assistants already cite.
Close the 6 readiness checks that did not pass
Starting with your unanswered reviews and the markup on your question pages; fixes for your developer, or done by us with access.
Earn a presence on the 9 sources the assistants cite for competitors and never for you
Most authoritative first. A listing, an article or a quote, each with your explicit yes.
Measured every week, done for you.
The audit tells you where you stand. The programme changes it. The same measurement runs every week, and the work it points at gets done: fixes to your site, content your buyers and the assistants can use, your Google Business Profile kept alive, and a presence on the sources the assistants trust, in the languages your buyers use. Searches like apartments for sale in your town now return actual listings inside assistant answers, so the programme also prepares and marks up the listings you choose, to give them their best chance of being the ones shown.
Measure
The same buyer questions every week, in every market and language you sell to, across every assistant in your tier.
Decide
What moved and what is missing becomes a ranked to-do list. The biggest gap is always next.
Draft
Briefs built from the pages the assistants already cite, drafted in your voice and checked by a person before you see them.
Approve
Nothing is published without your yes: one click from a weekly digest, with your edits or your veto.
Publish
On your site, on your Google Business Profile and on the outlets the plan chose, in the right language.
Verify
A week later we check whether the assistants picked it up and re-measure your mentions, so you see what worked.
Step 6 feeds step 1: every week starts from what the last week published.
Fresh numbers per market, assistant and question. One approvals digest each Tuesday. Nothing is flagged as a change until it has held for more than a week.
A written review in plain language: what moved and why, what was published, what comes next. Forward it to a colleague as it is.
On Authority: an executive review, a co-branded market report for your own marketing, and the quarter's placement targets agreed with you.
Reports carry the Propelos brand and are yours to share. Unbranded reports are available for multi-office groups.
Your name goes on the work, so nothing goes out unchecked.
Plenty of AI content services publish fast and fix later. Everything below exists because a wrong figure or an invented quote published under your name takes longer to repair than the piece took to write.
You approve everything, in two steps
Content for your own website and Google Business Profile arrives in a Tuesday digest, and you have five working days to approve, edit or veto each item with one click. Anything published under your name on someone else's site needs your explicit yes every time, and nothing goes out if you do not answer.
Verified figures only
A tax rate, a fee or a market figure reaches a draft only after we have found it at the authority's own source and read it there ourselves. Anything we cannot verify becomes a marked blank for your team to fill, never a guess.
Nothing invented
No made-up anecdotes, no invented reviews, no sources we have not read. A draft may only cite pages we actually opened. Where your own experience belongs, the draft leaves a clear space for it.
Human-reviewed, with no tell-tale AI writing
Every draft is read by a person and passes a per-language style gate that rejects the vocabulary, rhythms and punctuation that make AI text recognisable. This page went through the same gate.
Honest numbers, licensed data
When an assistant fails to answer, the report records a failed call, never your absence. Small samples are labelled as small. Question wording is frozen, so a number that moves reflects the market and never a reworded question. We use licensed APIs and a licensed data provider for the public ChatGPT and Gemini apps; we never scrape logged-in products.
Three tiers, one weekly measurement.
Every tier includes the weekly measurement, the approval workflow and the safeguards above, and each tier includes everything in the one before it. No setup fee.
We measure every week and tell your team exactly what to do next.
- 25 buyer questions tracked every week, in 1 market
- ChatGPT, Google AI Mode and Gemini
- Full audit with code-level fixes for your developer, plus a monthly advisory call
- Weekly dashboard and a monthly written review
- Competitor gap map, refreshed every week
- Content briefs for your team; Google Business Profile audit and post templates
- Alerts when a journalist, a forum thread or a published ranking is looking for an agency like yours
We write, publish and keep your local presence alive. You approve.
- 75 buyer questions tracked every week, in 2 markets
- All four assistants: ChatGPT, Google AI Mode, Gemini and Claude
- We implement the fixes wherever you give us access
- A step-by-step plan for every gap the measurement finds
- 3 pieces a month, drafted, human-reviewed and published with your approval
- 4 Google Business Profile posts a month, and your reviews answered
- Opportunity alerts with the answer already drafted
The full programme, with earned coverage on the sources the assistants trust.
- 150 buyer questions tracked every week, in 4 markets
- All four assistants, and your fixes first in our queue
- Quarterly executive review on top of the weekly dashboard
- Quarterly competitor deep dive
- 5 pieces a month, offered to partner outlets for republication
- 8 Google Business Profile posts a month and the profile Q&A filled, for every office
- We answer journalist requests and pitch for at least 2 earned placements a quarter (target 3); no outlet guaranteed
- €150 a month towards paid distribution, and a co-branded quarterly market report
No setup fee. Three-month initial term, then monthly rolling. Additional market +€200 a month; additional office +€150 a month. Prices exclude VAT.
Questions estate agents ask about GEO before they start.
What is GEO, and how is it different from SEO?
Is AEO the same as GEO?
Does GEO help my SEO, or compete with it?
Which AI assistants matter for property buyers?
Can you guarantee my agency will be named?
What do the assistants read before they name an agency?
Can my property listings themselves appear in AI answers?
How long until the numbers move?
Which languages can you work in?
Should we block AI crawlers to protect our listings?
What does my Google Business Profile have to do with ChatGPT?
Is AI visibility only for large real estate agencies?
Find out what the assistants say about you, before your next buyer does.
The free audit takes a few minutes and shows what the assistants say about your agency today, from fifteen buyer questions and searches in up to two languages. If you want the weekly programme after that, book a call and we walk through your report together.