What Schema Markup Does a Singapore Business Website Actually Need?

Use Organization to identify the company with a stable @id, its name, website, logo and official profiles. Add legalName, foundingDate and other identifiers only when accurate and supported by the business. Use sameAs for profiles that represent the same organisation, not unrelated directories or articles mentioning a similar name.

Use LocalBusiness or an appropriate subtype when it describes the actual business and location. Keep the published address and telephone consistent with the visible contact information. Do not invent a storefront, opening hours or coordinates. areaServed describes the service area; it is not evidence of a physical office.

A service page describes two related things: a document and a commercial offering. Give the WebPage its own @id, canonical url, name and inLanguage. Connect it to the relevant WebSite through isPartOf, and reference the Service through mainEntity.

Give the Service a separate @id with its name, description, serviceType, provider and areaServed. Its mainEntityOfPage can point back to the WebPage. Keep webpage attributes such as inLanguage and isPartOf on the WebPage rather than putting them directly on Service. Use the current page URL, not an unrelated service or a different language version.

For bilingual pages, check that each page has a self-referencing canonical URL and reciprocal hreflang links. Keep the company identity consistent across languages while giving each translated document its correct URL and language. Schema does not replace canonical or hreflang annotations.

Publish FAQs when they answer genuine buying or implementation questions. Answers about scope, customer responsibilities and limitations can be helpful even without special search presentation. FAQPage remains a Schema.org vocabulary type, but it should not be presented as a mandatory requirement for every business website.

Google's documentation updates state that FAQ rich results stopped appearing in Search on May 7, 2026. Do not sell FAQ markup as a route to that discontinued feature. Google also does not document FAQPage as a trigger that guarantees an AI citation. If you retain FAQPage markup for other consumers, its questions and answers should match the visible page.

Use Article on editorial content with a truthful author, publication date and modification date. A substantive correction should update dateModified while preserving the original publication date. Do not refresh dates automatically on every request. BreadcrumbList should describe the actual navigation hierarchy using working canonical URLs.

A testimonial is not a licence to promise review stars. Google has specific supported types and review policies; self-serving reviews of a LocalBusiness or Organization are not eligible for its review-star feature. Do not aggregate reviews from other websites merely to manufacture a rating on your own company page. Display genuine client feedback transparently and check eligibility before adding review markup.

Inventory the homepage, contact page, service pages and articles. List the facts visible on each template and choose matching types. Generate JSON-LD from the same source as the visible content so a CMS update does not leave stale structured data behind.

Inspect the deployed HTML for both languages. Parse the JSON-LD, verify that referenced identifiers agree, and confirm that URLs resolve to the intended pages. Use the Schema.org validator for vocabulary checks and Google's Rich Results Test for the search features it supports. A valid Schema.org type does not necessarily have a Google rich-result feature.

Use Search Console URL Inspection to check Google's view of important pages after deployment. Keep sitemap lastmod accurate. Record changes and compare relevant search impressions, clicks and enquiries over time; a before-and-after change alone does not prove that schema caused the result.

For a repeatable AI visibility check, record the platform, model when available, date, location settings, exact prompts and cited URLs. Separate brand mentions from linked citations and distinguish retrieval failures from a page that was retrieved but not selected. Repeat the same prompt set rather than relying on one favourable answer.

This guide does not claim that a particular percentage of AI-cited pages contains structured data, or that correlation proves a ranking or citation benefit. An earlier version contained inconsistent percentages and duplicated passages. Those claims and passages have been removed; the guidance now relies on the primary documentation linked below.

Google: AI features and your website

Google: Organization structured data

Schema.org: Service

Schema.org: WebPage

Google: documentation updates and FAQ rich-result retirement

Google: review snippet policies

Google: localized versions of your pages