The LocalBusiness schema was installed in Taiwan but was not mentioned by the AI engine. The problem is probably not in the program code, but in the misalignment of the data. When a user asks Perplexity or ChatGPT "Are there any recommended dentists in Xinyi District, Taipei?" AI will first cross-reference the name, address, and phone number on your website with Google business profiles, maps, and industry directories. As long as three versions appear in these places, it will tend not to mention you, rather than guessing the right one for you. So what really determines whether you will be cited is the consistency of NAP (name, address, phone). Schema syntax just packages this fact into a machine-readable format. This article goes through the implementation sequence: select the type, fill in the address, write the business hours, and finally align the NAPs of all platforms.
Why AI engines are particularly sensitive to LocalBusiness
When the AI engine answers local questions, it needs facts that can be extracted cleanly: where is the store, whether it is open now, and what is the phone number. Chinese addresses are not friendly to machines. "No. 7, Section 5, Xinyi Road, Taipei City" and "No. 7, Section 5, Xinyi Road, Xinyi District, Taipei City" are the same place in human eyes, but they may be two entities for the extraction model. LocalBusiness schema uses fixed fields to separate and mark this information, which is equivalent to directly handing the answer to the engine, saving it the cost of guessing and comparison. The less the engine needs to guess, the more it dares to include you in the answer.
Step 1: Choose the right @type, don’t stuff it all with LocalBusiness
There are dozens of subtypes of LocalBusiness under schema.org. The more specific you are, the better the engine will know what you do. An accounting firm is labeled as LocalBusiness, which is generally used, and it is labeled as AccountingService. There is a big difference in the AI's judgment of "whether this is the professional service I am looking for." First find the subtype closest to your business. If you really can't find it, then go back to the previous level and use a generic type.
- Restaurants and coffee shops: Restaurant, CafeOrCoffeeShop
- Dentist, Clinic: Dentist, MedicalClinic
- Lawyers and land agents: Attorney, LegalService
- Accounting and bookkeeping firm: AccountingService
- Retail store: Store, which can be subdivided into ClothingStore, HardwareStore, etc.
- Beauty, hair salon: BeautySalon, HairSalon
- Only when there is really no corresponding project will the general-purpose LocalBusiness be returned.
Address field: Map Taiwan administrative region to PostalAddress
The address is the most commonly written error. PostalAddress has fixed fields. Taiwan's administrative hierarchy must be aligned one by one. Don't stuff the entire series of house numbers into streetAddress, otherwise the engine will not be able to extract counties, cities and administrative districts.
- addressCountry: Fill in the country code "TW", not "Taiwan" or "Taiwan"
- addressRegion: municipality or county, such as "Taipei City" and "New Taipei City"
- addressLocality: district or town, such as "Xinyi District"
- streetAddress: road, section, lane, number, such as "No. 7, Section 5, Xinyi Road"
- postalCode: 3+3 postal code, such as "110011"
- geo: latitude and longitude, the coordinates must be consistent with the location of the pushpin on Google Maps
Business hours: openingHoursSpecification How to write without making mistakes
OpeningHoursSpecification is used for business hours, dayOfWeek is used to mark each period, and opens and closes are marked with HH:MM in 24-hour format. For shops that are closed at noon, don’t write it as 09:00 to 21:00, but split it into two sections (for example, 11:00–14:00 and 17:00–21:00), otherwise the AI will tell the user that you are open at 3 p.m. Don't list it if it's a public holiday. If it's open 24 hours a day, fill in 00:00 for both opens and closes. For temporary adjustments such as the Lunar New Year or national holidays, use specialOpeningHoursSpecification and add date coverage, which is safer than changing the main time directly. After making the change, remember to synchronize it with the holiday business hours of Google merchants. Both sides must say the same thing.
NAP consistency: schema is just one of the nodes
No matter how beautifully written the schema is, all your efforts will be wasted as long as other parts of it don’t match up. The name, address, and phone number should be the same verbatim on each platform, including punctuation and formatting. The phone number is especially confusing: the official website ends with "02-1234-5678", the Google business page ends with "(02)12345678", and the schema ends with "+886-2-1234-5678". It works for everyone, but only three pieces of information for machines. It is recommended that structured data should always use the E.164 format "+886212345678". The version displayed for human viewing can be typeset separately, but only one type is left in the schema.
- Google Business Profile
- The schema, footer and contact page of the official website
- Facebook, Instagram business pages
- Apple Maps and various navigation platforms
- Industry Directory and Local Yellow Pages
When we audit local official websites for clients, the most common flaw is not missing the schema, but the same phone number being written in four ways on four platforms. To the AI engine, it is not one store, but four stores that look very similar.— Tenten GEO audit notes
Pre-launch verification checklist
- Use Google Rich Results Test to confirm that the schema can be parsed and there are no errors
- Use Schema Markup Validator to check the syntax of each field
- Paste the NAP in the schema verbatim next to the Google merchant and compare it to catch any inconsistencies.
- Just ask the AI engine a local question and see if the address and business hours it answers are correct.
- Return visits regularly after going online. Once the address or phone number changes, all platforms will be synchronized.
The difficulty with LocalBusiness schema has never been that piece of JSON-LD, but keeping the facts scattered across a dozen platforms consistent over time. If you’re not sure how many stores your NAP and structured data count in the eyes of the AI engine, Tenten GEO’s 30-day GEO audit will identify these gaps all at once. If you want to quickly get directions first, you can also make an appointment for a 30-minute GEO diagnosis. We will use your actual website address to run a local query for you.
