Google's site: operator restricts a search to a domain, URL or URL prefix. Write site:example.com installation guide, with no space after the colon, to look for relevant material within that site. It is useful for finding documentation, exploring a section and spotting unexpected pages. It is not a complete inventory of Google's index.
Google's documentation says an indexed URL is not guaranteed to appear in a site query. Results without an additional search term also do not have the usual ranking meaning: the shortest URL generally appears first, with other results relatively random. The examples below use the reserved example.com domain to explain syntax; they are not reports of searches we performed or measured time savings.
Five core ways to use site:
| Search scope | Example | How to read it |
|---|---|---|
| A domain | site:example.com installation | Look for relevant results across the domain, including its subdomains |
| A particular subdomain | site:docs.example.com installation | Focus on the documentation host |
| A URL prefix | site:https://example.com/guides/ installation | Limit the URL scope to the specified prefix |
| A particular page address | site:https://example.com/guides/setup | Obtain a clue about whether that address is indexed and served; inspect the returned URL |
| A file format | site:example.com filetype:pdf installation | Find matching PDF material, then check its origin, version and applicability |
Start with the broadest scope appropriate to the task. Add the known host or directory when there is too much unrelated material. Google explicitly notes that site:https://www.example.com and site:https://example.com/ can return different results. A typo, an old domain or an unintended path can exclude the page you wanted.
A complete page address is still supplied to a prefix-based search operator; it is not an equality test in a URL database. Open the result and compare the actual address. For a site you manage, use URL Inspection when you need the stored indexing state of one specific URL.
Google's operator overview explains that filetype: uses the content type or file extension. A matching PDF is not automatically a current manual, an authorized copy or an original research paper. The file must still be read and checked.
Seven combinations for practical research
1. Add an exact phrase when ordinary keywords are too broad
Try site:example.com "installation clearance" when a particular phrase matters. Google documents quotation marks as an exact-match search option. This can help locate a phrase used in a manual or an older article.
Use it carefully: a relevant page might say “required clearance” instead. If the results are sparse, remove the quotes or try another term. The absence of one wording does not prove the site lacks information about the subject.
2. Exclude an irrelevant word
site:example.com bearing -skateboard is a syntax example for a search where skateboard material is distracting from an industrial-bearing task. The minus operator excludes a word from the query's results. Add exclusions one at a time and review what remains.
It does not identify advertisements, genuine buyers or trustworthy reviews. A query such as -sponsored is a word exclusion, not a reliable way to remove every paid promotion from a marketplace. It can also omit a useful page that happens to mention the excluded word.
3. Try excluding a host or directory
site:example.com -site:blog.example.com installation can help focus a search away from a blog subdomain. For a directory, try site:example.com -site:https://example.com/blog/ installation. Inspect the returned URLs to confirm the intended scope rather than assuming a complicated expression worked perfectly.
If the aim is simply to find product documentation, selecting its known host or prefix directly is often easier to explain and repeat than combining many exclusions. Begin with the positive scope you actually need.
4. Combine the site with a date range
Google's search-help documentation describes before: and after: as filters for documents last updated before or after a date. For example, site:example.com release after:2025-01-01 before:2026-01-01 illustrates a date-limited search for release information.
Open the selected page and verify the relevant date and version. Google's date guidance explains that it estimates publication or significant-update dates from several signals. The result is not a crawl-date log, proof of first publication, or necessarily the date of the event discussed. You cannot turn these results into a complete competitor publishing calendar.
5. Compare the same topic on two known sites
Run site:example.com material selection and site:example.org material selection separately. Save the query and the useful result URLs for each, then compare the actual answers: intended audience, specifications, supporting evidence and next steps.
This is a manageable starting point for competitor website analysis. Search-result counts do not establish a site's size, content quality or market share. A domain suffix alone also does not identify an author's location or certify expertise; choose the real organization and verify who produced the material.
6. Find a document with both a format and a distinctive phrase
site:example.com filetype:pdf "installation clearance" combines the file format with a more precise subject. Use the same approach with a real manufacturer domain and the exact model or series name when looking for a manual.
Before using a parameter, check the document title, model, revision, units and operating conditions. A newer brochure for a related product may not supersede the old manual for your model. Do not combine maximum values from different tables into a specification that no source actually states.
7. Look for unexpected content on your own site
Google lists site queries containing unexpected spam terms as one way to investigate a website. A query such as site:example.com casino can produce leads when that subject should not belong on the site.
Read the context before treating a result as a compromise: a legitimate article might discuss the term. If it is suspicious, preserve the URL and page evidence, then examine the site, access logs and Search Console security information. No matching result does not establish that every file and page is clean.
Why site-result counts are not an indexing report
Retrieval and display limits mean that a site query can omit indexed URLs. Google also notes that a narrower prefix can return more results than a broader one. Changing the query, host or path changes the observed set. The approximate number shown with a search should therefore not be used as an exact indexed-page total or a percentage of pages lost from the index.
For your own property, the Page indexing report helps investigate overall indexing categories, while URL Inspection provides information about a specific URL. Check the recorded result, selected canonical and crawl information. A live test examines present access and possible indexability; it does not mean Google has already indexed the changed page.
If your real problem is a submitted URL that is still missing, follow the indexing troubleshooting steps. Keep discovery, crawling, indexing and visibility for a particular query separate. A sitemap submission or a passing live test cannot stand in for all four.
A repeatable research record
For each useful search, record the exact query, date, language and available location settings, along with the result URLs you actually opened. Note the document version and what each source supports. If a page is inaccessible, record that limitation instead of treating the search snippet as its full contents.
For example, a manual search can yield a shortlist of documents; your output should identify the applicable model and revision after reading them. For content planning, it can reveal candidate pages to compare; your output should explain the useful answer or missing detail. The value comes from the verified material, not from claiming that a search operator guarantees a particular accuracy percentage.