Queries such as “a three-bedroom home in Guishan, Taoyuan, at or under NT$15 million,” “a rental within walking distance of National Central University in Zhongli for no more than NT$15,000 a month,” or “a three-bed, two-living-room place in Xinyi District under NT$50 million” show up every day in agency and rental work. They are also where smart search can help most.

Ask a local agent and the usual reply is that the homes are already on a major rental portal. If the agency’s own site never turns each listing into parseable structured data, those homes simply never enter the recommendation pool. A complete tailored site plan should treat every property as its own data record.

If you want smart search to keep matching your listings, a complete RealEstateListing schema is the necessary starting point. The steps below can be applied directly. For the broader site structure, see the property site planning notes.

RealEstateListing markup sample

Five setups worth doing

1. Map each property type

Listing kind	Schema type
Multi-unit housing or a tower	`Apartment`
Townhouse or villa	`House`
Office or shop	`CommercialProperty`
Factory or industrial building	`IndustrialBuilding`
Studio or a single rental room	`Apartment` + `accommodationCategory: Studio`

Pick the right type first, or later classification will drift. For how to fill the fields, see the Schema.org markup notes.

2. Write address and coordinates clearly

The more specific the location, the easier it is to surface the listing for “what is nearby” questions:

streetAddress: write the street number plainly; do not blur it
addressLocality: district
addressRegion: city or county
geo.latitude/longitude: keep coordinates to four decimal places

3. Do not leave spec fields blank

Buyers and renters usually check these details first:

numberOfRooms: room count
numberOfBathroomsTotal: total bathrooms
floorSize: area, including unitText
yearBuilt: construction year, used to infer age
floorLevel: floor and total floors
numberOfFullBathrooms: full bathroom count
amenityFeature: parking, lift, and similar features

4. Price and availability

"offers": {  "@type": "Offer",  "price": "36500000",  "priceCurrency": "TWD",  "priceValidUntil": "2027-06-30",  "availability": "https://schema.org/InStock"}

After a sale, switch the status to SoldOut so smart search learns the home is no longer open.

5. Action markup for viewings

Use potentialAction: ReserveAction to connect an online booking:

"potentialAction": {  "@type": "ReserveAction",  "target": {    "@type": "EntryPoint",    "urlTemplate": "https://example.com/listing/456/reserve",    "actionPlatform": ["http://schema.org/DesktopWebPlatform", "http://schema.org/MobileWebPlatform"]  }}

Listings that can be booked online usually rank better in recommendations.

Advanced: add registered-price context

  Area price context
  Sample annual average for Da'an District: about NT$1.26 million per ping (replace with the actual period from the interior ministry price register)
  Illustrative unit price for this listing: about NT$1.22 million per ping (about 3% under that area average)

Publishing open government figures gives a price context people can check, and smart search is more willing to cite it. Replace the sample numbers with the period and average you actually looked up.

Planning sketch for a mid-size team

The sketch below uses a mid-size team with about 200 listings to show which metrics to track after a full schema rollout. Swap in your own measured figures.

Metric	Before	After
Smart recommendations for the core area plus filters	[baseline]	[target]
Organic traffic to listing pages	[baseline]	[target]
Online viewing bookings	[baseline]	[target]
Average close rate after a booking	[baseline]	[target]

What actually separates results: one URL per home, a complete schema, a link to registered prices, and online viewing bookings.

Four pitfalls owners often hit

❌ No URL of its own: if every home sits on one page, smart search cannot recommend a single listing.
❌ Price written only as “call for details”: without a number, budget-based queries cannot match the home.
❌ Fewer than five photos: the listing looks thin, and booking intent usually drops.
❌ Sold homes left online with no status change: recommending a closed deal, then telling the caller it is gone, damages trust.

Closing: accurate data beats layout

Sale and rental sites are not mainly about visual polish. They depend on precise specs and a clear structure. Recommendations match conditions, not how pretty the page looks.

When planning an agency site, schema completeness should come before a visual refresh, and it often matters several times more. Local teams around Taoyuan and Zhongli frequently see listing-page organic traffic recover more from complete markup than from another ad budget. If registered prices, bookings, and client records also need to connect, see the custom system build option.

ARTICLE

Structured Markup for Sale and Rental Listing Sites: Helping Smart Search Recommend Properties More Accurately