How accurate is the site: operator

What the operator is

site: restricts Google results to a domain, a URL prefix or a single URL. Google's Search Central documentation lists it as a debugging aid for site owners and gives examples: site:example.com "returns a list of indexed and serving URLs", and site: with an exact page URL "may help you understand whether a specific URL is indexed and served."

It is the first tool most people reach for, and the method most often over-trusted. This page sets out where it is accurate and where it is not, so that the step in checking whether a URL is indexed is read correctly.

What Google says it does not do

The same documentation has a short limitations section. Two points matter for index checking:

1. Incomplete lists. The operator "doesn't necessarily return all the URLs that are indexed under the prefix specified in the query." Google specifically warns against using it to work out how many URLs are indexed.
2. No ranking. A site: query without search terms "doesn't rank the results". Google says it generally puts the shortest URL first, but otherwise the order is "relatively random".

Google also notes the operator "was designed primarily for search users", not as an index audit tool. Its Search Console help for the Page indexing report still suggests site: searches for small sites, under about 500 pages, as a quick way to see a sample of indexed pages.

Accuracy by result type

The asymmetry is the key fact. A hit is close to proof; a miss is a question. This is the main reason query-based tools add more query forms, as described in inurl: and quoted-URL checks.

Why indexed URLs go missing from site: results

Google does not publish a full list of reasons, so the following are documented behaviours that plausibly contribute, not an official explanation:

• Canonical substitution. If Google indexed a different URL as canonical, the checked URL may not appear while its canonical does. See canonicalization and index checks.
• Prefix mismatch. site:https://example.com/page and site:example.com/page/ are different strings. A protocol or trailing-slash difference changes what matches.
• Result trimming. For large prefixes, Google's documented position that the list is incomplete means a page can be indexed and still not shown.

Differences like these explain much of why index checks disagree between tools and between runs.

Using site: correctly for a single URL

1. Use the full URL exactly as published, including protocol.
2. Search from a private window to limit personalisation.
3. If it appears, record the result and stop.
4. If it doesn't, try the version without protocol, then with and without a trailing slash.
5. Still nothing: run a quoted-URL search and an inurl: search.
6. Owners: confirm with URL Inspection, which reports the indexed status directly.

Using site: for a whole domain

Domain-level site: queries are good for spotting things, poor for counting. Useful patterns from Google's documentation include site:example.com plus a spam term to spot hacked content, and site: with a topic word to see which pages can rank for it. For counts, owners should use the Page indexing report, whose totals Google describes as complete and accurate from its perspective.

Implications for checkers

A checker that runs only site: will report more false negatives than one that combines query forms. That is a design choice worth asking about when comparing tools for bulk index checking. The IndexChex checker, published by the owner of this site, combines site: with three other forms for this reason. Accuracy figures are only meaningful if the tool states how it measured them; see index checker accuracy and the glossary definition of site:.