PublishedVerifiedEvery 6 weeksSources8 namedAuthored bySquareRank Team
Boutique Realtors · AI search · § 4.12.1 · The wedge
Realtor AI Search Citations on Squarespace
When a buyer opens ChatGPT or Perplexity and asks "find a realtor who specialises in 1920s Craftsman bungalows in Bishop Arts" or "agent who works with first-time VA-loan buyers in Tacoma", the engine names three or four agents in its answer. Whether one of them is the boutique on your block depends on five things — and none of them are the Zillow8 Premier Agent profile that wins on the head term. AI engines cite a different set of pages on long-tail neighbourhood plus property-type queries, and the install that ships those signals (RealEstateAgent schema, a knowsAbout neighbourhood vocabulary, a 134-167 word answer-first lead per page) is the install that puts a single boutique agent in the citation card.
This leaf is the wedge across the boutique-realtor cluster. The neighbourhood + property-type query is precisely the shape Search Engine Land's 2026 GEO research1 shows AI engines absorb first, and it is precisely the shape a portal agent profile cannot answer well. The install changes are not novel AI magic — they are the same five-step framework every Pillar 1 page describes, applied to a vertical where named-neighbourhood vocabulary is the discovery currency and entity recognition lives on the individual credentialed agent, not the brokerage's logo.
§01The query shape
The neighbourhood + property-type query AI engines absorb first
Generic 'realtors in [city]' and 'homes for sale [city]' are Google queries — Zillow, Realtor.com, and Redfin own them with portal-grade domain authority, and the SEO contest there is settled. The queries that move in AI search are the ones with embedded neighbourhood, property-type, and buyer-segment constraints: 'historic district agent Oak Cliff', 'Craftsman bungalow specialist Bishop Arts', 'mid-century modern realtor Palm Springs', 'first-time homebuyer agent VA loans Tacoma', 'relocating military family agent San Diego'. These queries are too narrow for a portal profile's field set to answer fluently, and they are precisely the shape AI engines absorb most aggressively in 2026.
The shift is mechanical. Search Engine Land's 2026 GEO research1 documents that AI engines lift answers from passages matching the user's full constraint set, not just the head term. A page that addresses "Craftsman bungalow agent Bishop Arts" with a named neighbourhood ("Bishop Arts" — not "Oak Cliff" or "Dallas"), a named architectural style ("Craftsman bungalow" — not "older home"), a named credentialed agent (with REALTOR® use correct), and one specific local transaction detail beats a generic "find a real estate agent in Dallas" article on a major portal because the latter answers the head term but not the constraints. The engines reward the specificity.
The reason this is a real wedge for boutique agents — not just a theoretical opportunity — is the unusual density of valid constraint vocabulary in real estate. A buyer query commonly stacks neighbourhood, architectural style or era, property type (single-family, townhome, condo, multi-family), buyer segment (first-time, relocating, downsizing, investor), and a financing constraint (VA loan, FHA, jumbo, cash). The compounding effect is that the long-tail vocabulary in real estate is much larger than in most local-service niches, and the number of pages that can earn a citation for any one constraint stack is much smaller than the head-term contest implies.
The intent-shaped query landscape for boutique agents
~25%
projected drop in traditional search volume in 2026 as AI engines absorb intent-shaped queries. Hyperlocal real estate queries are squarely in the absorbed set.
What ChatGPT and Perplexity actually cite for realtor queries
ChatGPT and Perplexity cite pages they can reach, pages with passages that answer the user's full neighbourhood plus property-type constraint stack in 134-167 words, pages with an entity-recognised credentialed agent, and pages whose schema confirms the entity is a real practitioner at a real brokerage. The four filters compound. A boutique site can hit one filter and miss the others, and the citation lands on the portal listing instead. The job of the AI install is to clear all four filters on the same page in the order the engines apply them — crawler access first, passage shape second, entity recognition third, schema confirmation fourth.
The first filter is crawler access. Squarespace's AI exclusion checkbox6 ships unchecked by default, but a non-trivial share of brokerage sites toggled it on after 2024-era "protect listing data" advice — advice that conflated training crawlers (which scrape content into model training sets) with retrieval crawlers (which fetch a page in real time so the engine can cite the agent). The exclusion box treats both as one, so toggling it on blocks the citation path entirely. The first audit step is verifying the box is unchecked and that ChatGPT-User, OAI-SearchBot2, and PerplexityBot3 can all reach the bio page, the neighbourhood pages, and any active-listing pages the brokerage wants cited.
The second filter is passage shape. AI engines prefer pages where the first 134-167 words under each H2 directly answer the section's question without requiring a click — the format Search Engine Land calls "answer-first, expand for context". A boutique agent's bio targeting "Craftsman bungalow agent Bishop Arts" should open with a bolded passage naming the agent (with REALTOR® mark correct), the neighbourhood, the architectural style, and one specific transaction detail that proves real local fluency. Below the lead the page expands with depth — the historic district context, the named lending programmes that suit the area, the school catchment, the agent's transaction pattern in the neighbourhood over the past three years. The lead is what the engine quotes; the expansion is what the engine reads to confirm the agent is real.
§03The content
The 134-167 word answer shape that wins realtor citations
A citation-target bio or neighbourhood page for a boutique agent opens each H2 with a bolded one or two sentence lead, between 134 and 167 words, that names the agent (with REALTOR mark correct and credentialed designations spelled in full once), the named neighbourhood, the named architectural style or property type, the named buyer or seller segment, and one specific local transaction detail proving the agent's real local fluency. Below the lead, expand with the historic district context, named lending programme detail, school catchment specifics, recent transaction pattern, and any named publication credit the agent has earned. The lead is what the engine quotes; the expansion is what the engine reads to verify the agent is real and the brokerage is a real practice.
The contrast with a Zillow profile is the strategic point. A portal profile is field-driven and tag-led, so the prose on the profile is sparse — name, brokerage, sale count, areas served as a list of city names, a short bio. AI engines can read all of it, but they cannot extract a fluent constraint-matching passage from a tag set. A Squarespace bio page that puts the same information into a 134-167 word answer-first lead reads as a citable passage and the engine quotes it; the same information distributed across the portal profile's fields reads as a database row and the engine cites the portal instead.
Three details inside the example carry the citation. Specific named-neighbourhood vocabulary ("Bishop Arts", "Oak Cliff", "Davis Street", "Eighth Street historic corridor") — not "South Dallas" or "the area". Specific architectural and era vocabulary ("1920s Craftsman bungalow", "pier-and-beam foundation") — not "older home". Named lending programme specifics ("FHA 203(k)", "Texas Bond Program first-time buyer loan") — not "loan options". The engines reward this specificity because the user's query asked for it, and the page that supplies it directly is the page the engine has the highest confidence quoting.
§04Entity
The knowsAbout array and the named-neighbourhood vocabulary
For most local-service businesses, the entity AI engines need to recognise is the brand. For boutique brokerages, the entity that decides citation is the credentialed agent plus the named neighbourhood and property-type vocabulary the agent actually works in. The knowsAbout property on the Person schema is the canonical place to list the agent's real specialism vocabulary — named neighbourhoods, named architectural styles, named property types, named buyer segments — and AI engines read that array as the agent's specialism list. Without it, the agent is anonymous in the engines' entity graph; with it, the engines can confidently attribute a neighbourhood plus property-type citation to a real credentialed practitioner.
The knowsAbout property4 on a Person schema accepts a freeform array of subjects the person is knowledgeable about, and AI engines read it as the canonical list of named specialisms an agent is known for. The discipline that makes the array work is honesty: list the neighbourhoods the agent has actually closed transactions in and the property types the agent actually specialises in, not the brokerage's full MLS coverage area. The narrow honest list outperforms the wide aspirational list in citation share because the engines cross-check the array against the page content and the verified credentials, and an array that does not match what the page demonstrates is downweighted.
JSON-LDAgent Person schema with knowsAbout and the REALTOR mark, paste into the bio page Page Settings > Code Injection
<script type="application/ld+json">{"@context":"https://schema.org","@type":"Person","@id":"https://bishoparts-realty.com/team/jordan-lee/#agent","name":"Jordan Lee, REALTOR®","jobTitle":"Broker Associate","url":"https://bishoparts-realty.com/team/jordan-lee/","worksFor": {"@type":"RealEstateAgent","name":"Bishop Arts Realty","@id":"https://bishoparts-realty.com/#brokerage"},"hasCredential":"Texas Real Estate Commission License #0654321 · ABR · SRES · NAR member","knowsAbout": ["Bishop Arts historic district","Oak Cliff Craftsman bungalows","Kessler Park Tudor revival","First-time homebuyer guidance","Texas Bond Program financing","FHA 203(k) renovation loans","City of Dallas Conservation Districts"],"sameAs": ["https://www.linkedin.com/in/jordan-lee-realtor","https://www.realtor.com/realestateagents/jordan-lee_dallas-tx","https://www.zillow.com/profile/jordan-lee-bishop-arts/"]}</script>
The sameAs links matter on a credentialed practitioner schema. AI engines use them to disambiguate one named agent from another, and an agent with LinkedIn plus a Realtor.com profile plus a Zillow profile plus a state board listing is significantly easier to confidently attribute than an anonymous name. Linking to the portals as sameAs is counter-intuitive — the portals are competitors on the head term — but on the AI citation surface those links function as verification paths confirming the agent is a real listed practitioner at the brokerage they claim. The NAR-credentialed Realtor.com profile7 in particular reads as a trust signal the engines weight more heavily than a generic LinkedIn profile alone.
§05Measurement
Measuring AI citation for a boutique agent
AI citation measurement for boutique agents works the same way it does for any local-service business: a tracking spreadsheet of 10-15 neighbourhood plus property-type queries, run weekly across ChatGPT, Perplexity, and a Google AI Overviews trigger, with a column for whether the agent appears and a column for which other sources cite alongside. GA4 referrer data captures only a fraction of AI traffic (most arrives without a referrer header), so the manual log is the primary signal. Expect 6-12 weeks for the first visible movement after a full install, with Perplexity moving fastest because its citation card structure rewards named-credential pages, and AI Overviews moving slowest because Google's broader local-pack logic still tilts toward the portals on local queries.
The query list is the leverage. Generic queries ("realtors in Dallas", "real estate agent Dallas") are easy to track but unlikely to move in year one — the portal floor on those terms is too settled. Neighbourhood plus property-type queries ("Bishop Arts Craftsman bungalow agent", "Kessler Park Tudor revival realtor", "Oak Cliff first-time homebuyer agent") are harder to track because they require specific phrasing — but those are the queries where citation movement actually shows up. The right list mixes one or two slightly broader queries with eight to twelve constraint-stacked queries the agent genuinely wants to be the answer for. Update the list quarterly as the agent's transaction mix evolves and new neighbourhoods or property types enter the agent's vocabulary.
