A local AEO lab: documenting the Elda-Petrer business center pilot
We are testing AEO on a real local business and documenting the process from technical crawl access to entity signals and external sources.
We are starting a local AEO lab with Centro de Negocios Elda-Petrer, a business center in Elda that offers private offices, coworking, meeting rooms and company domiciliation.
The goal is not to publish a polished success story after the fact. The goal is to document the work while it happens: what we audit, what we change, what gets blocked, which sources matter and how AI engines respond over time.
Why this is a useful first pilot
Local AEO is measurable because buyer prompts are specific. People ask for coworking in Elda, private offices in Elda, meeting rooms near Petrer or a company address in Elda. The answer engines need clear entity facts and trusted local evidence to recommend a provider.
The initial baseline
- The website already had metadata, LocalBusiness schema, a sitemap, visible FAQ content, llms.txt and ai-facts.json.
- Cloudflare Managed Robots was injecting contradictory rules for AI crawlers; we disabled the managed robots setting and purged robots.txt.
- The contact form still needs a real sending path.
- The sitemap only contains the home page and legal pages, so service-level local pages are needed.
- The new center needs stronger external entity signals in maps, coworking directories and local business sources.
The local AEO method we are testing
The pilot treats local AEO as an evidence system, not as a campaign built from isolated keywords. For each service, the site needs a page that answers the buyer's question directly, a structured-data layer that describes the real entity, and external profiles that repeat the same facts. Google Search Central explains that structured data helps systems understand page content and gather information about companies and other entities. Schema.org defines LocalBusiness as a physical business or branch, which fits a local business center better than a generic organization description.
The working rule is simple: every fact an answer engine might reuse should be visible, consistent and correct in more than one place. The business name, address, service area, opening hours, phone number, services, booking path and evidence of real-world presence should match across the website, the Google Business Profile, map listings, directories and any local sources that describe the center. Google Business Profile guidance also emphasizes accurate location information, so this is not only an AEO preference; it is basic local search hygiene.
Evidence matrix for the pilot
- Entity facts: canonical business name, address, phone, legal or commercial identity, category, service area and sameAs profiles.
- Service facts: coworking, private offices, meeting rooms and company domiciliation, each with a clear page, eligibility, pricing or quote path where available, and contact route.
- Crawler facts: robots.txt, sitemap, llms.txt and ai-facts.json should not contradict one another or hide the pages that matter.
- Structured facts: LocalBusiness and service-level schema should reflect visible page content, not invented claims or hidden FAQs.
- External facts: map profiles, local directories, coworking listings and partner pages should use the same name, address, phone and service language.
- Measurement facts: prompt samples, engine responses, cited sources and visible changes should be stored with dates so movement can be interpreted over time.
How we will measure local answer visibility
The prompt set will start with buying questions, not vanity terms. Examples include "best coworking in Elda", "private offices in Elda for a small team", "meeting room near Petrer" and "where can I domicile a company in Elda". Each prompt will be sampled repeatedly across answer engines. The useful output is not one screenshot where the business appears; it is a dated record of whether the center is mentioned, whether the site or an external profile is cited, which competitors appear, and which facts the answer gets right or wrong.
This matters because local answer engines often combine web pages, map data, business profiles and third-party sources. If the center is absent from a response, the next question is not "how do we force a recommendation?" The better question is "which evidence did the engine trust instead, and what fact is missing from our public footprint?" That framing keeps the work practical and avoids the false promise that any agency can guarantee a specific AI answer.
What would count as progress
- The service pages are indexed and internally linked from the home page and sitemap.
- The answer engines can identify Centro de Negocios Elda-Petrer as a real local business with specific services.
- Prompts start returning correct facts even when the center is not recommended.
- External profiles and directories stop contradicting the website.
- Citations, mentions and competitor sources can be compared against a stable baseline instead of anecdotal checks.
Limits of the pilot
This pilot cannot prove that one edit caused one AI answer to change. It also cannot guarantee inclusion in ChatGPT, Gemini, Claude, Perplexity or Google AI experiences. Local AEO is probabilistic: the inputs can be improved, the evidence can be made cleaner, and the measurement can show whether visibility is improving, but the final answer is controlled by each engine.
What we will do next
The next implementation work is not generic blogging. It is entity and source work: service pages, crawler files, sitemap updates, maps profiles, external directories, review signals and repeated prompt measurement.
- Create pages for coworking, private offices, meeting rooms and company domiciliation in Elda.
- Update robots.txt, llms.txt, ai-facts.json and structured data so answer engines can extract facts cleanly.
- Build external presence in Google Business Profile, Bing Places, Apple Business Connect and relevant coworking directories.
- Track prompts such as "coworking in Elda", "private offices in Elda" and "where can I domicile a company in Elda" across answer engines.
AEO becomes more useful when it is documented as a working system, not sold as a finished theory.