Start with the property report, not an ordinary search query
To investigate links Google reports for your site, open Search Console in your own authorized account and select the intended property. A search for your domain in ordinary Google results is a different task. It can reveal pages to investigate, but it does not produce a complete backlink inventory. The question here is where a reported reference came from and what evidence to save.
Check whether you selected a Domain property or a particular URL prefix. Record that selection before comparing numbers or deciding that another host is external. If the property is unavailable, ask its owner for the appropriate access. Google’s permissions table includes Links report access for restricted users; a full user is not simply another name for read-only access.
Navigate from a target page to a linking page
Open Links, then the external-links section. For a destination-first investigation, open the expanded top-linked-pages table, select your target and then a linking site. The final table lists source pages. Alternatively, start with top linking sites, choose the publisher and then the target page. The official report guide documents both paths. Save which route and filters led to the record.
Read the scope before interpreting counts. Targets are grouped by canonical URL, sites by root domain, and duplicates are combined. The report is a sample, includes references found over time and does not report nofollow status. A historical row may remain after the reference changes. These are reasons to inspect a selected source, not reasons to assume a complete live list.
Work through a fictional navigation log
Our invented shop wants to inspect references to a printable planner. The log below is an authored exercise, not a screenshot of a Google account or a report on a customer. The selected property is a URL-prefix property. A source on a publisher subdomain appears under its root-domain group; that grouping does not mean the root homepage is the exact linking page.
The example’s target-difference field contains an old planner URL. The investigator separately follows that URL and observes a redirect to the selected planner. That redirect is invented case evidence, not something the report alone proves. Another real case could involve canonical grouping instead. Preserve the reported URL and the observed route rather than overwriting one with the other.
| Step | Fictional selection or record | What to save | What it does not prove |
|---|---|---|---|
| Select property | https://shop.example/ | URL-prefix scope and access context | Coverage of every protocol or subdomain |
| Choose target | https://shop.example/planner | Selected canonical target | Exact href on every source |
| Choose publisher group | publisher.example | Root-domain group | A link on the root homepage |
| Open linking page | https://news.publisher.example/guide | Exact source URL and inspection date | Current presence without inspection |
| Keep the different target | https://shop.example/old-planner → https://shop.example/planner | Reported old URL and separately observed redirect | That every target difference is a redirect |
| Export with context | Latest links; discovered 2026-10-20 | Export type, collection date and raw file | Publication date or complete inventory |
Choose the export for the question
For a focused investigation, export the expanded single table and save the chosen target and publisher in your notes. Google documents a 1,000-row table limit. On the report overview, the latest-links and broader-sample exports each have a maximum of 100,000 rows. Latest means recently discovered by Google, not necessarily recently published by the source. More sample links is still a sample.
Name the raw file with property, export type and collection date. Keep it unchanged and work from a copy. Record visible missing-data markers: Google notes that unavailable values may appear as zeros in downloaded reports. Do not reinterpret an unavailable measurement as a verified absence. The worksheet here records decisions; it is not a substitute for the actual export.
Verify a selected reference and retain its context
Open the source page, locate the reference and record the actual destination, link text, complete relationship attributes and observation date. If access fails, leave the page result unresolved. A report row and a current page inspection answer different questions. The companion checking guide explains this source-selection and inspection workflow in more detail.
Use the completed CSV to follow the six steps, then use the blank log for your authorized property. Save the original export alongside the log, including any different target URL. These instructions do not connect your Google account. The documented Search Console API lists Search Analytics, Sitemaps, Sites and URL Inspection; it does not expose a Links report method. Search Analytics rows must not be presented as backlink records.
Download the worked report log · Download a blank report log
Keep search visibility and backlink evidence separate
Google’s site: operator documentation describes results from a specified site or prefix and warns that they are not exhaustive. A site-restricted result is not, by itself, an inbound reference. A domain mention may also be plain text rather than a link. Open the relevant page and inspect the element before recording a backlink.
A missing report row is not proof of deletion; a present row is not proof of current delivery, a follow attribute or ranking benefit. Compare like-for-like property and export scopes before investigating a change. This guide serves the report walkthrough, not named third-party tutorials or a promised all-web Google backlink query. Keep unresolved questions visible so another reviewer can reproduce the investigation.