Short answer: no tactic guarantees that an AI assistant will recommend your project. What is documented: each platform has a search crawler you can allow or block, Google says its AI features need no special markup beyond normal Search eligibility, and an academic benchmark found that some content changes can raise visibility in generated answers by up to 40% (an experimental result, not a promise). In practice, GEO is solid SEO plus verifiable data, clear structure and a consistent presence outside your own site.

Want to talk about your project? Message us on WhatsApp.

What is Generative Engine Optimization (GEO)?

GEO is the practice of making your content more likely to be used and cited when a generative search engine answers a question. The term was formalized in a paper by Aggarwal, Murahari, Rajpurohit, Kalyan, Narasimhan and Deshpande, accepted at KDD 2024, which defines generative engines as systems that gather information from multiple sources and summarize it with language models (arXiv:2311.09735). You will also see “AEO” (answer engine optimization) and “AI search optimization.” There is no single industry definition.

Why does GEO matter for real estate developers and investors?

Because the first research step for many overseas investors is now a question typed into an assistant: who builds in a given market, which projects are in pre-sale, what the buying process looks like for a foreign national. Two data points on the direction of travel:

• Pew Research Center followed 900 U.S. adults in March 2025. Google users clicked a regular result link on 8% of visits when an AI summary appeared and on 15% when it did not; about 1% clicked a link inside the summary itself (Pew, July 22, 2025). This is an observational U.S. sample, not a measure of any specific market.
• The GEO paper reports visibility gains of up to 40% in generative engine responses and notes that effectiveness varies by domain (arXiv). Treat that as an upper bound from a benchmark, not an expectation for your site.

If readers get the answer without clicking, being named in the answer, or listed as a source, becomes part of brand visibility.

How do the major assistants access your website?

Providers do not publish ranking formulas. What they do document is which crawlers exist and what they are for:

The decision to allow search crawlers and the decision to allow training crawlers are separate. A developer who wants visibility in AI answers typically allows the search crawlers and decides on training access (GPTBot, ClaudeBot, Google-Extended) as a policy question with counsel.

What technical fixes matter most on a project website?

1. Make the content readable as text. Google advises keeping important content in text form and allowing crawling in robots.txt and at the CDN or host level (Google). A project site that shows everything inside images, scanned PDFs or a 3D viewer with no text is effectively blank to these systems.
2. Check your firewall. OpenAI and Perplexity both note that, beyond robots.txt, requests from their published IP ranges should be allowed. A “bot protection” rule at your CDN can block them without any visible error.
3. Give every project a fact page. Name, location, unit types, size ranges, construction status, estimated delivery, developer, last-updated date and contact. This is the page an assistant is most likely to lean on for factual questions.
4. Keep structured data consistent with the page. Google says markup you use must match visible content, and that no special markup exists for AI features. Schema.org does define relevant types, such as ApartmentComplex, with properties like `numberOfAccommodationUnits` and `tourBookingPage`. Add it for clarity, not as a ranking trick.
5. Publish an English version. If your target is U.S. and international investors, a properly written English page (not a machine-translated afterthought) gives assistants an English source to cite. This is a general practice, not something the platforms document as a requirement.
6. llms.txt is optional. It is a proposal by Jeremy Howard (September 2024) for a Markdown file that summarizes a site for language models. The official page still describes it as a proposal and does not claim models read it natively (llmstxt.org). Do not treat it as a requirement.

A minimal robots.txt that allows the search crawlers (adjust with your technical team):

“` User-agent: OAI-SearchBot Allow: /

User-agent: Claude-SearchBot Allow: /

User-agent: PerplexityBot Allow: / “`

What kind of content gets cited?

The GEO paper built a benchmark of diverse queries paired with web sources and reported visibility gains of up to 40% for its methods, with results varying by domain (arXiv). The abstract does not rank individual tactics, and lists of “winning tactics” in marketing blogs often come from secondary summaries. What follows are sensible practices consistent with Google’s guidance on helpful, reliable, people-first content, not a proven recipe:

• Lead each page with a direct answer to the question the page targets (“What does a foreign buyer need to purchase pre-construction property in the Dominican Republic?”).
• Attribute numbers. State the figure, the source and the date, and link to the primary source (central bank, statistics office, published statute).
• Define terms briefly. Investors from other markets do not share vocabulary on trusts, pre-sale contracts or title.
• Use comparison tables for unit types, phases and requirements.
• Show authorship and update dates.
• Publish what only you know: real construction timelines, floor plans, the questions investors actually ask you, your land selection criteria.

Avoid padding pages with statistics to look citable. An unverifiable number is a reputational liability, and it can be repeated by an assistant long after you correct it.

What matters beyond your own website?

Google notes that its AI features may run several related searches across subtopics and data sources (“query fan-out”), which surfaces a wider set of links than classic search (Google). That points to the surrounding ecosystem:

• Consistent company details (name, address, phone, description) across Google Business Profile and industry directories. Google recommends keeping Business Profile data current.
• Coverage in trade media and portals with consistent project facts.
• LinkedIn and YouTube presence with technical material aimed at investors.
• Verifiable references from delivered projects, used only if real and authorized.

How do you measure GEO results?

There is no standard metric yet. Practical options:

1. Quarterly prompt audit. Write 20 to 30 questions your investor would ask (“developers in Punta Cana,” “pre-sale projects near Cap Cana,” “how foreign buyers purchase in the Dominican Republic”) and run them in ChatGPT, Gemini, Claude and Perplexity. Record whether you are mentioned, what is said and which sources are cited. Repeat with the same script each quarter.
2. Search Console. Google states that AI Overviews and AI Mode traffic is reported in the Performance report under the “Web” search type, combined with other web traffic (Google).
3. Referral analytics. Look for visits referred from AI assistant domains.
4. Server logs. Confirm which crawlers visit and whether they receive errors.

AI answers vary from run to run, so look for trends rather than a single screenshot.

Should an international investor-facing site be written differently?

Probably, yes. Investors reading in English ask different questions from local buyers: how title works for foreign nationals, what the closing process looks like, how funds move, who the developer is and what it has delivered. Answer those on dedicated pages with dated facts and primary sources, and have counsel review anything about legal or tax treatment before publishing. Platforms do not document a language-specific tactic, so this is judgment about matching content to the questions people ask.

What mistakes should you avoid?

• Promising “guaranteed ranking in ChatGPT.” Nobody controls those answers.
• Mass-publishing unreviewed AI-written pages, which lowers quality and trust.
• Inventing awards, rankings or sales figures to look authoritative.
• Leaving stale information (prices, availability, delivery dates) online. An assistant may repeat it.
• Blocking all bots by default and then wondering why AI tools never mention you.

GEO checklist for a project

• [ ] robots.txt and firewall allow the search crawlers you chose.
• [ ] Key content is HTML text, not only images or PDFs.
• [ ] Project fact page is complete and dated.
• [ ] Each important page opens with a direct answer.
• [ ] Every figure has a linked source and a date.
• [ ] Structured data matches visible text.
• [ ] Company profile is consistent across Google and directories.
• [ ] Quarterly four-assistant prompt audit is documented.

Frequently asked questions

Does GEO replace SEO?

No. Google states that its AI features have no additional technical requirements and no special markup, and points to standard Search fundamentals. GEO adds citable content discipline and a crawler review on top of sound SEO.

Can I pay to be recommended by ChatGPT?

There is no public, reliable mechanism for organic answers. Be skeptical of anyone selling it as a guarantee. What you control is whether your site is accessible, clear and verifiable.

Should I let AI companies train on my content?

That is a business decision. Search crawlers and training crawlers are controlled separately (for example OAI-SearchBot versus GPTBot at OpenAI), so you can allow one and block the other.

How long until changes show up?

No timeline is published. OpenAI and Perplexity say robots.txt changes can take about 24 hours to be reflected; the effect on answers depends on reindexing and competition for the question.

Is llms.txt worth publishing?

It is cheap to publish, but it remains an open proposal without confirmation that models read it natively. It does not replace the steps above.

What if an assistant gets my project wrong?

Publish the correct, dated information on your own site, fix third-party pages that repeat the error, and use each platform’s feedback channel. Keep the project fact page as your source of truth.

Need help?

Netix builds project websites and landing pages with SEO, CRM integration and commercial content, along with market studies and investor dossiers for real estate projects in the Dominican Republic. See Netix’s real estate marketing and commercialization services (page in Spanish).

Informational; not legal or technical advice. Crawler policies and AI features change often; confirm them in each platform’s documentation. Figures as of October 13, 2026.