176 of the 178 AI-Ready websites publish Organisation schema. The tag is worth 1.45 points.

RBy Ronnie Huss, founder of SearchScore

In July we registered a question on SearchScore Labs before running it: do sites with Organisation schema score higher on the entity and authority sub-score than sites without it? The registration committed us to publishing the answer either way, including a null.

The answer is yes, and by a wide margin. Sites carrying Organisation schema average 34.56 on that sub-score against 18.73 for sites without it, a gap of 15.83 points with a 95% confidence interval of 15.61 to 16.04, measured across 33,034 sites with the markup and 69,840 without.

First, a correction to our own registration

The registration assumed that sub-score was independent of schema. It is not, and that is worth saying before anything else. Seventy-seven of its hundred points sit on checks a JSON-LD Organization block can help satisfy, because the scorer appends the block’s own sameAs list to the page HTML before matching social and profile URLs, and a PostalAddress is one of the routes to the address check. Read straight off that metric, the result would be partly circular.

So we ran the test the registration should have specified. Twenty-three of the hundred points sit on checks JSON-LD cannot touch under any circumstances: whether the site has an about page, a contact page, a privacy policy, a terms page, a newsletter signup, and established-domain indicators. If the association were an artefact of the scorer reading its own schema, the gap would collapse on those.

The gap is larger on the points schema cannot reach. Sites with Organisation schema are 25 points more likely to publish a privacy policy and 27 points more likely to link a contact page. No JSON-LD produces that. The finding survives the correction, and survives it more convincingly than it would have passed unexamined.

The gap is twelve times larger than the tag

Organisation schema is one check worth 10 of the 83 points in the structured data category, which carries a 0.12 weight in the score. Publishing it moves your overall result by 1.45 points. The sites that have it score 17.77 points higher. If the schema block were doing the work, those two numbers would be close.

That shows up most clearly in the categories that contain no Organisation schema check at all.

A site that publishes Organisation schema is, on average, 39 points better at platform optimisation and 29 points better at topical authority. Whoever added the fifteen lines of JSON was also the sort of person who set the canonical tags, wrote the headings properly and gave the site a subject.

The gradient is close to absolute

Read from the top, that looks like a requirement. Read from the bottom, it looks like a symptom. Both readings are supported by this data, and the difference between them is the whole practical question.

The one place the advantage nearly disappears

Across five categories the schema group leads by between 15 and 39 points. On AI citability it leads by 3.73.

AI citability measures whether an engine can reach your content, read it and safely quote it: crawler access, structured citations, quotable statistics, an llms.txt. It is the part of the score most directly tied to being used in an AI answer, and it is the one place where the well-built sites are barely ahead.

This is correlational and we are not going to dress it up. The design compares sites that already differ in many ways. It cannot tell you what happens when a given site adds Organisation schema tomorrow, and 1.45 points is the only part we can promise. A controlled follow-up, measuring the same sites before and after they add it, is the experiment that would settle it.

Squaring this with our own index

The SAVI Q3 report says these signals are now load-bearing. This piece says one of them is worth 1.45 points. Both are correct, and the gap between them is the useful part.

The report describes a class of signals in aggregate, and the data here agrees with it: sites carrying them score 17.77 points higher. What this experiment adds is the mechanism. The load is not carried by the tag. It is carried by whatever kind of operation publishes the tag, and the tag is the cheapest way to see that operation from outside. Read that way the report’s advice stands unchanged. What this data does not support is the assumption a reader is likely to bring to it, that adding the markup transfers the advantage.

The United Kingdom is ahead, and it helps less

Among the 5,482 GB domains in the panel, 45.2% publish Organisation schema against 31.8% across the weighted panel. The effect is slightly smaller there: 14.65 points on the entity sub-score against 15.83 globally. The signal is more common in the UK and worth a little less, which is what you would expect from a marker as it becomes ordinary.

What to do with this

Publish the schema. It takes fifteen lines, it is a genuine 1.45 points, and 176 of the 178 sites that cleared the AI-Ready bar have it, so its absence is conspicuous. Our schema markup guide covers the implementation. Do it in an afternoon and stop thinking about it.

Then do not expect it to have changed anything, and go and work on the things the schema group is only 3.73 points better at. Crawler access, quotable statistics, structured citations and content an engine can lift a clean sentence from. That is where the AI-Ready sites are not meaningfully ahead of everybody else, which makes it the part of the field still open. You can see where your own site sits with the free audit.

Method

Source is the SearchScore SAVI index panel, wave 1, dated 1 August 2026: 102,873 domains, all audited on scorer version geo-145-2026-07 with method quick+geoOnly, weighted to the population. Cells are 33,034 sites with Organisation schema and 69,840 without, against a pre-registered minimum of 300 per cell. Confidence intervals are 95%, computed on weighted means using Kish effective sample size.

The registered primary metric was the entity and authority sub-score, which the scorer calls brand authority. Point values come from the scorer: it totals 100 points, of which 77 sit on checks a JSON-LD Organization block can help satisfy and 23 sit on page-structure checks it cannot reach. Both subsets are reported above. That 1.45 points is what the check itself is worth, derived the same way: Organisation schema is 10 of the 83 points available in the structured data category, and that category carries a 0.12 weight.

Detection accepts Organization, Organisation and the LocalBusiness subtypes in JSON-LD, plus microdata itemtype equivalents. Country is taken from the site’s own declared markup, so the GB cut is sites that identify as GB rather than sites resolved by IP. Note that in quick mode, which is how the panel was scored, the Wikipedia check resolves only from a wikidata sameAs in the markup rather than an API lookup, so it fires on 2.6% of schema sites and 0.8% of the rest and contributes little either way.

To confirm the extract sits on the same basis as the published index, we recomputed the figures the SAVI Q3 report states. Weighted mean score reproduces at 45.24 against 45.24 published, Organisation schema at 31.8% against 32.4%, JSON-LD at 44.3% against 45.7%, llms.txt at 27.3% against 27.6%, FAQ schema at 3.4% against 3.4% and author bios at 15.9% against 16.1%. Our measurement standards set out what these numbers are fit for.

Pre-registered 8 July 2026 as EXP-0006, before the query was run, with a commitment to publish a null. The hypothesis was confirmed on the registered metric and on the stricter subset. The interpretation offered above, that the signal is a marker rather than a lever, goes beyond what the registration asked and is stated as interpretation, not as measurement.