The properties Google recommends and almost nobody uses

Key findings

• 103 of the 206 properties Google recommends are used by fewer than 100,000 websites.
• Counted as recommendations rather than properties, the gap is 109 of 314, 34.71%.
• Merchant listing alone has 18 of its 44 recommendations thinly used, more than any other feature.
• Google has retired 11 rich result features, and FAQ is one of them.
• I read the requirements from Google's own pages on 11 August 2026, not from memory.

Counted as properties the gap is 50.00%, counted as recommendations it is 34.71%

The table runs both counts through the same four stages, from what Google asks for to the share that is thinly used, side by side.

One feature asking for one property. A property asked for by nine features counts nine times. Distinct Schema.org properties, each counted once however many features recommend it. Both shares are correct, and they answer different questions about the same gap. I published the wrong one first, and the method page says how. If you quote one, the distinct-property share is the one I would quote, because it does not count a property nine times for being asked for by nine features.

Retail features carry 40 of the thinly used recommendations

July 2026 · six ranges, no exact numbersGoogle and the Schema.org community, Schema.org usage statistics dataset. Websites in Google's index.

Each bar is one feature's recommendations, filled by how many point at a property used by fewer than 100,000 websites.

Retail is the largest single cluster, though not the majority of the gap. Merchant listing, return policy and shipping policy account for 40 thinly used recommendations between them, and the same cluster shows up from the opposite direction in the usage file, where 22 terms Schema.org has not yet ratified are already on 100,000 websites or more, mostly merchant returns and product variants. Read together, those two facts say the same thing. Retail markup is moving faster than either Google's recommendations or Schema.org's own approval process. I did not expect a single cluster to stand out this clearly. If you run a shop, the recommended properties on these three features are where your markup is most likely to be thinner than Google's page suggests it should be.

Google has retired 11 rich result features

The table lists every removed feature with the date it went and Google's own stated reason.

Structured data advice ages badly, and a lot of what you will still find online recommends markup for features that no longer exist. FAQ is the one most often got wrong. You will read that it is now restricted to health and government sites. It is not restricted. It is gone, and the markup you add for it does nothing. I still find FAQ markup on sites built this year, which is why this table is here at all. If your own markup predates a date in it, that row is the one to check first.

Known limits of this data

• Google lists some things in its requirement tables that are not Schema.org properties at all. Twenty-five of them are vacation rental amenities such as hotTub and wifi, and one is xPath on the speakable feature. I keep them here because Google asks for them, and none of them carries a usage figure, because there is nothing in Schema.org for them to have one against.
• Names with a dot in them, such as offers.price and ListItem.position, are Google notation for a property nested inside another type rather than Schema.org terms in their own right. The usage figure against them is for the last part of the name.