Real estate SEO has been taught the same way for fifteen years: target the city, write area guides, collect citations, earn backlinks, keep the site fast. None of that is wrong, and all of it still matters. What has changed is that a growing share of property searches never produces a list of blue links at all.
A buyer asks an assistant for a three bedroom house near a good school with a short commute, and reads one synthesised answer built from a handful of sources. So the discipline now has two halves: the fundamentals that decide whether you are eligible to be found, and a newer layer that decides whether a machine can use what it finds.
This guide covers both, starting with the part most guides stop at.
What Real Estate SEO Still Requires
Nothing below is new, and skipping it makes everything after it pointless. A property site that cannot be crawled will not be cited by an assistant either, because assistants are grounded in the same index.
| Area | What matters most for property sites | Common failure |
|---|---|---|
| Indexation | Listing pages reachable by crawlable links, not only by search filters | Inventory behind JavaScript filters, invisible to crawlers |
| Site speed | Fast image heavy galleries, sensible lazy loading | 40 full size photos loaded at once |
| Local signals | Accurate business profile, consistent name, address, phone | Office address differing across directories |
| Content | Area and market pages that answer real questions | Thin duplicate suburb pages spun at scale |
| Authority | Local press, partnerships, genuinely useful tools | Bought links, expired-domain schemes |
| Expired listings | A deliberate policy for sold or withdrawn properties | Thousands of dead URLs, or mass deletion |
That last row is the one most portals get wrong and it costs more than it looks. When a property sells, the page still holds accumulated authority and still receives search demand for the address. Deleting it wastes both, while leaving it live and unchanged tells search engines you are advertising something unavailable. Marking it as sold, keeping the address and neighbourhood context, and linking to comparable current listings preserves the value.
The fundamentals decide whether you are eligible. They no longer decide whether you win, because your competitors have them too. What follows is where the gap is currently open.
Where Real Estate SEO Changes in AI Search
Property portals invest heavily in content and still lose visibility in AI answers. The reason is structural, and we covered it in depth in why real estate listing pages hide in AI search.
A listing page is a data entity wearing the costume of an article. It contains price, floor area, room count, address, and a description, but presents all of it as styled text. Watch what happens to one sentence when a human and a machine read it.
| The listing says | A buyer understands | A retrieval system extracts |
|---|---|---|
| "Spacious three bedroom" | Room to live in | bedrooms: 3 (only if marked up) |
| "Quiet street" | Low traffic, restful | nothing |
| "Close to everything" | Amenities nearby | nothing |
| "Great schools nearby" | Good for children | nothing |
| "Easy commute" | Short travel to work | nothing |
Four of the five claims buyers care about most are invisible. Traditional advice does not fix this, because it assumes the unit of competition is a document. In AI search the unit of competition is a fact.
The Three Layers of a Listing That AI Can Read
| Layer | What it contains | Typical state |
|---|---|---|
| 1. Property | Price, size, bedrooms, bathrooms, year built, tenure | Present, but as text rather than markup |
| 2. Location | Distances, travel times, named amenities, transit, schools | Missing or written as adjectives |
| 3. Question | Plain answers to what buyers actually ask | Almost never present |
The property layer is what every listing already has. The work is not adding information but exposing it in schema.org markup. One detail matters more than people expect: use the type that actually describes the property. A condominium marked up as an apartment complex is a factual error a machine will confidently propagate.
The location layer is where most listings are empty, and where the competitive gap sits. Distance to the nearest station and how long that walk actually takes, schools within a defined radius, supermarkets, parks, healthcare. Not adjectives. Distances, counts, names, travel times.
The question layer translates the first two into the shape of real queries. Buyers do not search for "walkability index 78". They ask whether they can live there without a car.
What Buyers Ask That Listings Cannot Answer
| Buyer question | Answerable from data? | On a typical listing? |
|---|---|---|
| How far is the nearest primary school? | Yes | No |
| What is the commute at 8am? | Yes | No |
| Supermarket within walking distance? | Yes | No |
| Where is the nearest park? | Yes | No |
| Is there a station I can walk to? | Yes | No |
| How big is the kitchen? | Yes | Yes |
Every location question is answerable from data, and almost none are answered. So the assistant fielding that buyer's question sources its answer somewhere other than your listing. That is the whole game: the listing that answers the question gets cited, the listing that describes the kitchen does not.
This is the mechanism we measured in citation rates for listings with and without structured geo data, and the one behind Google Ask Maps ranking listings on attribute match rather than proximity.
Building the Location Layer
1. Start from coordinates, not the address string. An address is a label, and label matching is fragile across formats and languages. Geocode each listing once, store the coordinates, treat them as the anchor for everything else.
2. Query the surroundings by category and radius. Pull the points of interest buyers care about within a walkable distance, then a driving distance. Store names and distances, not summaries.
3. Convert distance into time. "800 metres to the station" is a fact. "A ten minute walk" is the same fact in the form the buyer asked for.
4. Expose it as markup and as prose. Structured data so a machine can parse it, readable text so a human can use it and an assistant can quote it. Doing only one wastes the work.
Here is the second layer added to a listing that already has the first:
{
"@context": "https://schema.org",
"@type": "SingleFamilyResidence",
"numberOfRooms": 3,
"floorSize": { "@type": "QuantitativeValue", "value": 118, "unitCode": "MTK" },
"geo": { "@type": "GeoCoordinates", "latitude": 52.3702, "longitude": 4.8952 },
"amenityFeature": [
{ "@type": "LocationFeatureSpecification",
"name": "Primary school within 600 m", "value": true },
{ "@type": "LocationFeatureSpecification",
"name": "Metro station, 8 minute walk", "value": true },
{ "@type": "LocationFeatureSpecification",
"name": "Supermarket within 400 m", "value": true }
]
}
Our GeoEnrich API returns this surrounding context from a coordinate instead of a dozen separate lookups, and property discovery covers the same ground from the search side.
If you would rather start from a worked example than build one, we maintain an open real estate geo guide on GitHub with a 25 field property listing schema, agent profile and open house examples, and a verification checklist. It is free, and it is the fastest way to see what a complete listing entity looks like before committing to your own.
Schema Mistakes That Quietly Cost Property Sites
Markup errors are worse than missing markup, because a machine treats a confident wrong statement as fact and repeats it. These are the ones we see most often on property sites.
| Mistake | What it looks like | Why it hurts |
|---|---|---|
| Generic type for everything | Every property marked Residence | Loses the distinctions buyers filter on |
| Condominium as apartment complex | ApartmentComplex on a single unit | Describes a building, not the thing for sale |
| Price without currency | "price": 385000 alone | Ambiguous across markets, often dropped entirely |
| Floor size without a unit code | "floorSize": 118 | 118 what? Square metres and square feet both plausible |
| Coordinates rounded too far | Two decimal places | Places the property up to a kilometre away |
| Markup that contradicts the page | Schema says 3 beds, copy says 4 | Trust signal lost, the whole entity gets discounted |
| Agent details on every listing | Same RealEstateAgent block repeated | Fine, as long as the agent is not marked as the property |
The coordinate one deserves emphasis. Two decimal places of latitude is roughly a kilometre of error, which is enough to move a property into a different school catchment or the wrong side of a station. Every location attribute you compute afterwards inherits that error, so an imprecise coordinate silently corrupts the entire location layer.
Area Pages: What Still Works and What Never Did
Neighbourhood and area pages remain among the strongest assets a property site can own, because they answer questions no individual listing can and they stay relevant when inventory turns over. They also produce more thin content penalties than any other page type in the sector, because the temptation to generate one per suburb at scale is enormous.
The distinction is whether the page contains anything a machine could not have generated from a template. A page that lists average prices, transport links, school names, amenity counts, and how those numbers changed over the last year is a genuine reference. A page that swaps the suburb name into the same three paragraphs is duplicate content wearing a different hat, and search engines have been reliable at spotting the difference for years.
Area pages are also the natural home for the location layer at neighbourhood scale rather than property scale. The same data that answers "is there a supermarket near this house" answers "what is the amenity density in this district", and the second question supports far more search demand than any single listing.
Measuring Real Estate SEO When Answers Replace Clicks
The uncomfortable part is that the familiar scoreboard gets less informative. A listing can be the source an assistant used and generate no click, because the buyer got their answer in the conversation.
| Metric | Still useful? | Why |
|---|---|---|
| Keyword position | Partly | Says less every quarter as answers replace links |
| Organic clicks | Partly | Undercounts answers where you were the source |
| Citation presence | Yes | Did assistants name you when asked buyer questions? |
| Attribute coverage | Yes | How many buyer questions can your page answer from data? |
Pick ten listings and ten questions buyers actually ask. Count how many of the hundred combinations your pages can answer from data on the page. Tracked monthly, that number forecasts AI visibility better than any rank tracker, and unlike rank it is fully under your control.
Our AI SEO checker tests how a page reads to an answer engine, which is a reasonable starting point.
A Real Estate SEO Checklist, In Priority Order
| Priority | Action | Why here |
|---|---|---|
| 1 | Fix crawlability of listing inventory | Nothing downstream matters if pages are not indexed |
| 2 | Fix markup on templates, not listings | A template error repeats across every property you own |
| 3 | Correct the property type in schema | Wrong types propagate as confident wrong facts |
| 4 | Set an expired listing policy | Recovers authority most portals throw away |
| 5 | Add the location layer to top listings | Measure lift before rolling out to the portfolio |
| 6 | Add the question layer as FAQ content | Depends on the data underneath being right |
| 7 | Continue local and authority work | Slower cycle, still compounding |
The fundamentals are table stakes and your competitors have them. The location data layer is the clearest differentiator currently available, and it is the one most portals have not built.
Real estate has always been about location. Search is finally catching up, and it wants that location expressed as data rather than as adjectives.
Frequently Asked Questions
What is real estate SEO?
Real estate SEO is the practice of making property listings and agent websites discoverable in search. It covers the technical foundation that lets search engines crawl and index listing pages, the local signals that connect an agency to the places it serves, the content that answers buyer questions, and the authority earned through links and reputation. Since 2025 it also covers a fourth area: whether a listing exposes machine readable facts that AI assistants can extract and cite when answering a buyer's question in natural language.
Does SEO still matter for real estate if buyers use AI assistants?
Yes, and more than before, because the same underlying signals feed both. AI assistants are grounded in indexed web content and structured data. A listing that is crawlable, schema marked, and rich in verifiable location attributes performs well in classic search results and is far more likely to be retrieved and cited by an assistant. What changes is emphasis: keyword density matters less, factual completeness matters much more.
What structured data should a property listing page include?
At minimum, use schema.org markup that describes the listing accurately. Choose the correct type for the property rather than defaulting to a generic one, since a condominium is not an apartment complex and the distinction affects how a machine reads it. Include address components, geographic coordinates, price, size, and number of rooms. Then extend beyond the property itself: nearby amenities, transit access, walkability, and school proximity are the attributes buyers actually ask about, and they are what an assistant needs to match a listing to a natural language question.
Why do property listings fail to appear in AI search results?
The usual cause is structural rather than editorial. Listing pages are data entities pretending to be articles. They present price, size, and location as styled text for humans, with nothing machine readable underneath, so a retrieval system has no reliable facts to extract. A second cause is thin location context: the page describes the property in detail but says nothing verifiable about the neighbourhood, which is precisely what a query like "family home within walking distance of a primary school" depends on.
How does location data improve real estate SEO?
Location data turns vague claims into checkable attributes. "Great location" is unusable to a retrieval system, while a verified distance to the nearest station, a walkability score, and a list of amenities within a defined radius are all matchable facts. Adding that layer to a listing widens the set of questions the page can answer, which increases both long tail search coverage and the likelihood of being cited when an assistant answers a location shaped question.
How long does real estate SEO take to work?
Technical fixes such as indexation and schema markup can show results within weeks, because they change how quickly and accurately search engines understand pages you already have. Local and authority work runs on a longer cycle, typically three to six months before movement is clear. Location data enrichment sits in between: the markup is read quickly, while the compounding benefit of covering more buyer questions accumulates as those questions get asked.
