No test shows structured data lifting AI citations. Here's what it does for apartment websites.

• Resources
• Get started

Resources · Your website

No test shows structured data lifting AI citations. Here's what it does for apartment websites.

A labeled record in your website's code tells search engines and AI exactly what your community is called, where it is and how to reach it. AI engines check that website often: in our measurements over the last 30 days, ChatGPT cited a community's own site in 49% of the answers that named it. Here is what the record does in AI search, what it does not, and how to add it so every AI engine can read it.

By HaloScore TeamUpdated October 2026 · 7 min read

ShareLinkedInEmailCopy link

✦AI answer

What's the address and phone number for Wickmere Apartments?

Wickmere Apartments is at 880 Ballast Street, Maple Heights, WA 98000.

The leasing office is at (425) 555-0142, and tours can be booked on its website.

Sourceswickmereapartments.comapartments.com

In wickmereapartments.com's code

"@type": "ApartmentComplex","name": "Wickmere Apartments","telephone": "+1-425-555-0142","streetAddress": "880 Ballast Street","petsAllowed": true

Illustration. Names are invented.

• Key findings
• HaloScore data
• The basics
• The research
• The takeaway
• How to do it
• Questions
• Sources

Key findings

• AI engines check your website. In our measurements over the last 30 days, ChatGPT cited a community's own website in 49% of the answers that named it.
• Most communities we check already have it. Structured data is a labeled record in the website's code stating the community's name, address and phone. 83% of the communities we have checked publish one, an early read we will update.
• It is groundwork. Google says it is not required for its AI features but recommends keeping it, and Microsoft says it may help Copilot base its answers on the right facts. The largest independent study found no lift in AI citations from adding it.
• Put it in the page's own code. Not through Google Tag Manager, and with the same facts visible as plain text on the page.

HaloScore data

ChatGPT cites a community's own website in 49% of the answers that name it

HaloScore puts the questions renters ask to the major AI engines every month and records every source each answer cites. Two patterns from those measurements frame this article.

AI engines check a community's own website. When an AI engine names a community, it often cites that community's own website as a source for what it says. Over the last 30 days, ChatGPT did so in 49% of the answers that named a community, Google AI Mode in 51% and Google AI Overviews in 42%.

The website matters most once a renter asks about you. On "best apartments in..." questions, property and management-company websites are 8% of the sources ChatGPT and Google's AI Overviews and AI Mode cite. On comparisons between two communities they are 25%, and on a specific fact, such as a deposit, pet policy or parking, 21%. Listing sites lead on every kind of question.

So the website is one of the places AI engines go to confirm a community's details. Structured data is the part of the website that states those details in a form a machine does not have to interpret.

Figure 1

AI engines often cite a community's own website when they name it

Share of answers that named an apartment community, last 30 days.

AI engine

Answers citing the community's own website

ChatGPT

49%

Google AI Mode

51%

Google AI Overviews

42%

Source: HaloScore measurements, September 12 to October 12, 2026.

haloscore.ai/resources

The data in this chart

AI engine	Answers citing the community's own website
ChatGPT	49%
Google AI Mode	51%
Google AI Overviews	42%

Figure 2

Listing sites lead, and property websites rise on questions about facts and comparisons

Share of the sources cited by ChatGPT, Google AI Overviews and Google AI Mode in answers about apartment communities, last 30 days.

Search questions ("the best apartments in...")

78%

Lifestyle questions ("pet-friendly apartments with a pool")

50%14%35%

Comparisons between two communities

51%25%20%

Questions about a specific fact

51%21%26%

Listing sitesProperty and management websitesReview sitesEverything else

Source: HaloScore measurements, September 12 to October 12, 2026. "Everything else" includes property websites we do not recognize, so the property share is a floor.

haloscore.ai/resources

The data in this chart

	Listing sites	Property and management websites	Review sites	Everything else
Search questions ("the best apartments in...")	78%	8%	4%	11%
Lifestyle questions ("pet-friendly apartments with a pool")	50%	14%	2%	35%
Comparisons between two communities	51%	25%	5%	20%
Questions about a specific fact	51%	21%	3%	26%

What we found

21%

of the sources cited on factual questions were property and management-company websites, alongside listing sites.

83%

of the communities we have checked publish a property-level record in their website's code.

How we measured: HaloScore's measurements of AI answers to renter questions about the apartment communities we measure, September 12 to October 12, 2026, each answer counted once. A website is counted only when we recognize it as a property or management-company site, so that share is a floor. Source shares leave out Gemini, which we ask to answer from Google Maps. The structured data check reads each community's homepage as it is first delivered, before any scripts run, for an ApartmentComplex or Residence record or a LocalBusiness record with the community's address. It is an early read, and we will update it as more communities are checked.

Key takeaway

Your website is one of the places AI engines confirm your details, so every fact on it, from the address to the pet policy, needs to be exact and current.

The basics

Structured data only helps when it describes the community

Every page a renter sees has code behind it. Structured data is a small block in that code that labels facts: this is the community's name, this is its street address, this is the leasing office's phone number. It is usually written in a format called JSON-LD (JavaScript Object Notation for Linked Data), using the shared vocabulary published at schema.org.

What is schema.org?

schema.org is a shared vocabulary for structured data, founded by Google, Microsoft, Yahoo and Yandex and developed in the open.^[1] It names the record types, such as ApartmentComplex, and the facts each type can carry, so a record means the same thing to every system that reads it.

schema.org has a record type made for this, ApartmentComplex, which can carry the name, address and phone along with the pet policy, amenities, number of units and a link to book a tour.^[2] Across the web, though, most structured data describes something else: the most common types on home pages are WebSite, Organization and WebPage,^[3] which describe the website or the company behind it, not a place a renter can live. Most of the communities we check do publish a property-level record. The ones that do not can still carry structured data that tells a search engine nothing about the property, so the question is not whether a site has structured data but what it describes.

Figure 3

Across all websites, the most common structured data describes the site, not a place

Share of desktop home pages carrying each schema.org type, 2025: all websites, not apartment websites. None of the top types describes a place a renter can live.

schema.org type

Share of desktop home pages

WebSite

37%

SearchAction

28%

Organization

27%

WebPage

25%

ListItem

21%

BreadcrumbList

21%

ImageObject

21%

Source: HTTP Archive, Web Almanac 2025, search engine optimization (SEO) chapter, figure 5.52.^[3]

haloscore.ai/resources

The data in this chart

schema.org type	Share of desktop home pages
WebSite	37%
SearchAction	28%
Organization	27%
WebPage	25%
ListItem	21%
BreadcrumbList	21%
ImageObject	21%

Key takeaway

Having structured data is not the same as describing the community: what counts is a property-level record, such as ApartmentComplex, with the community's own name, address and phone.

The research

Google and Microsoft say it helps their systems, but the largest test found no lift in AI citations

Google and Microsoft both say structured data helps their systems read a page. The independent tests are less encouraging.

Google updated its guide to its AI features in July 2026: "Structured data isn't required for generative AI search, and there's no special schema.org markup you need to add." It still recommends structured data as part of SEO, because it makes a page eligible for Google's enhanced search listings.^[4] Its general documentation says it uses structured data "to understand the content of the page".^[5]

Microsoft goes further, with a caveat. A Bing product manager wrote in October 2025 that schema "helps search engines and AI systems understand your content", and listed it among the ways to be included in Copilot's answers, which are built on Bing's search index.^[6] Bing's current guidelines add the caveat: structured data "may support clearer grounding but does not guarantee visibility or grounding traffic", and "markup must accurately reflect visible content".^[7]

When an AI assistant reads a page while answering, the tests so far show it working from the text a visitor can see. In two controlled tests, searchVIU's in late 2025 and Otterly.AI's in March 2026 across seven AI assistants, including Google's AI Overviews and AI Mode and Microsoft Copilot, no assistant used a fact that appeared only in a page's structured data.^[8][9]

Figure 4

No AI assistant used a fact found only in structured data

Otterly.AI's structured data test (March 23, 2026) and Writesonic's test of six assistants on one page (March 30, 2026). A dash means the assistant was not in Writesonic's test.

AI assistant	Ran the page's scripts	Used a fact found only in structured data
ChatGPT	No	No
Claude	No	No
Gemini	No	No
Microsoft Copilot	Yes	No
Google AI Overviews	-	No
Google AI Mode	-	No
Perplexity	-	No

Source: Writesonic, March 30, 2026 (scripts);^[10] Otterly.AI, March 23, 2026 (structured data).^[9]

haloscore.ai/resources

Does it raise citations? The largest controlled study we found, by Ahrefs in 2026, compared 1,885 pages that added JSON-LD with 4,000 similar pages that did not. Over the following 30 days it found "no major uplift in citations on any platform": a small decline in AI Overviews and no measurable change in AI Mode or ChatGPT.^[11] The study looked only at pages that AI was already citing heavily, and did not test apartment or local business records, so it is not the last word. It is the best evidence available.

Figure 5

Adding structured data did not lift AI citations on any platform

1,885 pages that added JSON-LD, against 4,000 similar pages that did not. Only the AI Overviews change was statistically significant; the other two were indistinguishable from zero.

AI engine

Change in citations

Google AI Overviews

-4.6%

Google AI Mode

+2.4%

ChatGPT

+2.2%

Source: Ahrefs, May 2026.^[11]

haloscore.ai/resources

The data in this chart

AI engine	Change in citations
Google AI Overviews	-4.6%
Google AI Mode	+2.4%
ChatGPT	+2.2%

The takeaway

Treat structured data as groundwork that states your facts once

No test so far has shown structured data winning AI recommendations by itself. It is groundwork, and it does three useful jobs:

1. 01It states your facts once, without room for interpretation, on a page the AI engines already check.Google uses it to understand pages and recommends keeping it, Microsoft says it may help Copilot base its answers on the right facts, and Google's 2026 guide to AI search for marketers recommends LocalBusiness markup for local businesses.^[12]
2. 02It records your version of the facts.When your website, your listing sites and your Google Business Profile spell the name or address slightly differently, the record on your own site states the version you consider correct, for any system that reads it.
3. 03It is the norm among the communities we check.83% of them already publish one, so a site without it is missing something most of its peers have.

How to do it

One record, in the page's own code, matching your listings

• One property-level record for the community, at least on the homepage: ApartmentComplex, or a LocalBusiness record for the leasing office.
• The same name, street address and phone that appear on your Google Business Profile and listing sites, character for character.
• Only facts that stay true. Pets allowed, amenities, number of units and a tour booking link are worth adding. Leave rents out and link your live availability page instead, so the record never needs updating by hand.
• The same facts visible as plain text on the page. Microsoft's guidelines require markup to match the visible content,^[7] and the AI assistants that read pages live rely on the text.
• In the page's own code, not added by Google Tag Manager. Google can read a record that Tag Manager adds.^[13] Many AI assistants cannot: in a March 2026 test, ChatGPT, Claude and Gemini read pages without running their scripts.^[10] A record in the page's own code reaches every reader.

What is Google Tag Manager?

Google Tag Manager is a Google tool for adding and managing tags, such as analytics and advertising code, on a website without changing the website's own code.^[14] The tags are added by a script as the page loads, so an AI assistant that reads a page without running its scripts never sees them.

A worked example

Wickmere Apartments

Wickmere's website platform added structured data by default, but view the page source and it describes only the website and the management company. Wickmere's marketing manager asks the website provider to add an ApartmentComplex record to the homepage template, using the name, address and phone exactly as they appear on Wickmere's Google Business Profile. The provider offers to add it through Tag Manager; the manager asks for it in the template instead. The result is a block like this one, in the page's own code:

{
  "@context": "https://schema.org",
  "@type": "ApartmentComplex",
  "name": "Wickmere Apartments",
  "url": "https://www.wickmereapartments.com/",
  "telephone": "+1-425-555-0142",
  "address": {
    "@type": "PostalAddress",
    "streetAddress": "880 Ballast Street",
    "addressLocality": "Maple Heights",
    "addressRegion": "WA",
    "postalCode": "98000",
    "addressCountry": "US"
  },
  "petsAllowed": true,
  "tourBookingPage": "https://www.wickmereapartments.com/tour"
}

Questions

Common questions

Our website provider says the site already has schema. Is that enough?

Check what it describes. Often it covers the website or the management company rather than the community. Open the homepage, view the page source, and search for "ApartmentComplex", or for a LocalBusiness record that carries your community's own name and address. If neither is there, the community is not described.

How do we check that the record works?

Paste the page address into the Schema Markup Validator at validator.schema.org, a general-purpose checker for any schema.org record.^[15] Google's Rich Results Test only checks the types that produce Google rich results, and ApartmentComplex is not one of them, so it can report nothing even when your record is correct.^[16]

A study says structured data lifts AI citations 21.6%. Is that wrong?

It measures something else. Semrush's 2026 study reports a 21.6% citation lift for "structured data elements", but its method section says it "focused solely on the text visible on the page" and did not evaluate "schema markup", so the figure describes how visible content is formatted, not the code this article is about.^[17]

Get started

See where your community stands in AI search

HaloScore checks for a property-level record on every assessment, along with whether your name, address and phone can be read from your site's code, and tracks both each month alongside how often the AI engines recommend you.

Get started with HaloScore

Useful? Share it with your team or your management company.

ShareLinkedInEmailCopy link

Sources

1. schema.org, home page: what schema.org is and who founded it (version 30.1, read October 9, 2026). schema.org
2. schema.org, ApartmentComplex (version 30.1, September 2026). schema.org
3. HTTP Archive, Web Almanac 2025, SEO chapter (published January 2026). almanac.httparchive.org
4. Google Search Central, "Optimizing your website for generative AI features on Google Search" (updated July 10, 2026). developers.google.com
5. Google Search Central, "Introduction to structured data markup in Google Search" (updated December 2025). developers.google.com
6. Krishna Madhavan, Microsoft Bing, "Optimizing your content for inclusion in AI search answers" (October 2025). about.ads.microsoft.com
7. Microsoft, Bing Webmaster Guidelines (read October 9, 2026). bing.com
8. searchVIU, "Schema markup and AI in 2025: what ChatGPT, Claude, Perplexity and Gemini really see" (December 2025). searchviu.com
9. Otterly.AI, schema markup test across seven AI search platforms (March 23, 2026). otterly.ai
10. Writesonic, "AI crawler study: what LLMs see on your website" (six AI assistants, one test page, March 30, 2026). writesonic.com
11. Louise Linehan, Ahrefs, "Does schema markup increase AI citations?" (May 2026). ahrefs.com
12. Google, SEO in AI search: a guide for marketers (2026, PDF). services.google.com
13. Google Search Central, "Generate structured data with JavaScript" (updated December 2025). developers.google.com
14. Google Tag Manager Help, "Introduction to Tag Manager" (read October 9, 2026). support.google.com
15. schema.org, "Announcing Schema Markup Validator: validator.schema.org" (May 2021). blog.schema.org
16. Google Search Central, Rich Results Test help: the result types it supports. support.google.com
17. Luke Harsel, Semrush, content optimization for AI search study (January 2026); see its method section. semrush.com

ResourcesGet started

support@haloscore.ai© 2026 HaloScore LLC. All rights reserved.