Search Console Links Report Limits: 1,000 Rows and Sampling

The Search Console Links report cannot tell you that a link does not exist, and it cannot tell you that a link is still live. Its tables stop at 1,000 rows, it describes itself as a sample, it keeps links that may have been removed, and it does not say whether a link is nofollow. Anyone reading it for backlink purchases, publisher reviews or agency reporting should treat each row as a lead to confirm, never as a final verdict.

For a walkthrough of opening and using the report, see checking backlinks with the Search Console links report. The focus here is narrower: what the report leaves out and how that changes the sentences you are allowed to write.

What the help page actually states

The Search Console Help page for the Links report (Opens in a new tab) is short, and the useful part is precise wording. Reading it closely avoids the habit of filling gaps with assumptions.

Stated limits

• Tables are limited to 1,000 rows, so tables may be truncated on larger or more linked-to sites.
• The report is not a comprehensive list of every link. It shows a sample, and some URLs, such as non-indexed pages or deduplicated URLs, may be omitted.
• The report does not specify whether a link is marked nofollow.
• Listed links may have since been removed, or the page may no longer exist.
• Data is grouped by root domain, and pages are grouped by canonical URL, with duplicates combined.
• Exports reach up to 100,000 rows, while single-table pages export up to 1,000 rows, and values shown as a tilde or dash are exported as zeros.

Not stated

The page gives no figure for how long a new link takes to appear, no description of how the sample is chosen and no statement about a link's effect on rankings. Where the documentation is silent, the accurate sentence is "the documentation does not say." Invented timelines like "give it two weeks" belong to opinion, not to the report.

The 1,000-row ceiling and the export

Where the cutoff bites

On a small site the limit never shows. On a large site, a table that ends at exactly 1,000 rows tells you that rows beyond it exist but are not displayed. The top of a sorted table is reliable as a ranking of what the report chose to show; the tail is not. A newly acquired link from a small publisher is exactly the kind of row that can fall in the tail, which is consistent with the limit and not evidence of a missing delivery.

What an export gives you

Exports can reach 100,000 rows. The page describes two variants: latest links, which covers the most recently discovered links, and more sample links, which is a sample drawn from the full set known to Google. Single-table pages export up to 1,000 rows. Because tilde and dash values become zeros in the file, any total you compute from an export should treat those cells separately. If you use the interface in another language, button and tab names may look different.

Narrow the question instead of widening the table

Practical suggestion: ask the report narrower questions. Open the linking sites for one target page, then the links from one site to that page. Narrow views are another way to look at the same sample; the documentation does not promise they reveal rows missing from the main table, so record the answer as sampled data either way.

The merged-row problem

What gets merged

Protocol, subdomains such as m. and www, and subdirectories are stripped, so a publisher's main site, mobile subdomain and a subfolder section all count as one root domain. Top-level domains stay, so the same name under different extensions is different. A duplicate is a link from the same source URL to the same target URL once parameters and similar differences are removed, and duplicates combine. In the text table, duplicates are case-insensitive exact matches of the link text, and links without text show as "(empty)".

Why this matters when you bought links

Two links from one publisher on two articles can appear as a single linking site. That is not a missing delivery. Open the detail view for that site and look for each source page that points to your target. Conversely, a publisher running separate domains may appear as separate rows, which is useful when you evaluate whether one owner operates a family of sites.

Nofollow: the missing column

What the report cannot say

Because the report does not specify nofollow, a row tells you nothing about whether a link carries a qualification. You need the source.

What the rel values mean

Google's documentation on qualifying outbound links (Opens in a new tab) describes three values. Sponsored marks advertisements and paid placements; the page notes nofollow remains acceptable there but sponsored is preferred. Ugc is for user-generated content such as comments and forum posts. Nofollow is a catch-all for when the other values do not apply. Several values can be listed in one rel attribute separated by spaces or commas, and links marked this way will generally not be followed, though the target can still be found through other routes such as sitemaps or links from elsewhere.

Matching rows to rel values

Practical suggestion: for every row you care about, open the publisher URL, find the anchor element, and write its rel value in your sheet. A purchased link with no qualification, or a link expected to carry sponsored that has none, is a separate finding. The report will never surface that difference.

Filling the gaps

The page source

It answers whether the link exists and whether address, anchor and rel match the order. It is the only source here that is both current and specific to the link.

Link tools

Tools run their own crawls and can include sites the report's sample leaves out. Their methods vary by vendor, so treat a tool row as a second lead, not as confirmation. For comparing tools and reading a whole profile, see how to read a backlink profile and use link checkers.

The Manual Actions report

The Manual Actions report (Opens in a new tab) shows whether the site has manual actions and their history. Per the help page, an unnatural links action means Google detected a pattern of unnatural, artificial, deceptive or manipulative links to the site. To resolve it, the page advises trying to get the links removed first, with disavow reserved for those you cannot remove, and warns that disavowing everything without attempting removal may lead to rejection. When no action exists the report shows a green check mark. Checking it separately matters because an incomplete link list cannot tell you either way.

Your own records

A dated order sheet with publisher URL, ordered target, anchor and the delivery-day source check is the record that survives every report limit. Put it next to the report export so rows that are "listed" and "verified" can be seen side by side.

Reading rules: say this, not that

Practical suggestion: run each sentence in your report through these rules.

1. If the sentence says a link is absent, is the proof the report or the page source? If only the report, write "not listed in the report."
2. If it says a link is live, was the row checked against the page? If not, write "listed in the report."
3. If it gives a count, did the table reach 1,000 rows, and did root domain grouping affect the count?
4. If it mentions nofollow or sponsored, did the value come from the page source? If not, delete the sentence.
5. If it mentions rankings or link value, delete it. No column in the report shows that.

Example scenario: a cycling accessories shop

Example scenario: A cycling accessories shop buys three links from travel and fitness blogs. At month end the "top linking sites" table shows two of the three blogs. The owner could tell the seller the third link is missing. Instead the owner opens all three publisher pages: each carries the link, two with rel sponsored and one with no rel at all. The third blog's absence from the report is consistent with the documented limits, so it is logged as "not listed, verified in source." The missing rel value becomes a separate conversation with the seller.

Report sentences that hold up

• "Not listed in the report, which is not comprehensive; confirmed in the page source on the check date."
• "Listed in the report; no link found on the publisher page, so the row may be historical."
• "Table ends at 1,000 rows; the tail was not evaluated."
• "No rel data in the report; the source shows sponsored."

The anchor text table has the same limits

The "top linking text" table is where many people look for anchor patterns, and every limit above applies to it. It is capped at 1,000 rows, it is sampled, and it counts duplicates as case-insensitive exact matches of the text, so "Blue Widgets" and "blue widgets" are one row. Links with no text appear as "(empty)", which is what you should expect from an image link whose alt text is missing, since Google uses the alt text of an image link as its anchor.

What you can and cannot read from it

You can see which phrases appear in the sample. You cannot see how many distinct links sit behind each phrase in the full set, and you cannot see which of them were paid for. A phrase that looks over-represented may simply be over-represented in the sample. For planning anchors in the first place, the guide on anchor text strategy for link profiles is the better starting point; the report is a place to compare that plan against what Google has listed, with the caveats of a sample.

A cautious reading routine

Practical suggestion: note the top phrases, then open three or four of the publisher pages behind them and read the anchor in context. If the anchors in the live pages match your orders and read naturally, the table adds little worry. If a phrase appears that nobody ordered, trace it to its source page before deciding anything.

Misreadings that cost money or time

Paying twice for a link that exists

A buyer who trusts a missing row, tells the seller the link is absent and orders a replacement may end up with two links where one was already delivered. Opening the page first costs a minute and avoids the duplicate order.

Accepting a link that is gone

The reverse happens when a buyer sees a row and signs off. If the row is a historical record and the link was removed, the acceptance is based on a stale line. A source check on the day of sign-off closes this gap and creates a dated record.

Escalating over a merged row

A buyer who orders two links from one publisher sees one linking site and accuses the seller of under-delivery. The merge rule explains the single row. A detail view or the source pages settle the question without a dispute.

Mixing sampled months

Comparing this month's export of "latest links" with last month's export of "more sample links" compares different slices. Keep the export type constant when you compare periods, and say which type you used.

Reading a clean report as a clean bill of health

A tidy-looking table says little about manual actions. The Manual Actions report exists for that purpose, and the documentation presents a green check mark there when no action exists. Looking at both reports closes a different gap than looking at one twice.

How the limits land on different readers

Buyers

The buyer's question is whether the paid link is live. The report answers it indirectly: a row may be stale, absent or merged. The delivery decision therefore rests on the publisher page.

Publishers

The report describes links pointing at your own property, so it does not list the outbound links a publisher gives. A publisher answering a client's question about a link placed on their site should consult their own page source and content records.

Agencies

An agency that merges reports risks comparing unlike things: one property may hit the row ceiling, another may be defined at a different level. Note property type, export type and row counts for each client in the report.

Small business owners

A business owner often asks for one number. The row count is not "links earned," so report it as "the report lists this many linking sites; it is not a complete list," and pair it with the number of deliveries verified in the source.

Troubleshooting when a link is not in the report

1. Open the publisher URL. If it does not return a successful page, the problem is the page, not the link.
2. Look for the anchor element in the source. If it is absent, delivery is incomplete and the report is irrelevant.
3. If present, compare href, anchor text and rel to the order.
4. If all match, look for your target page in the top linked pages table and then for the publisher domain in its linking sites.
5. If the table hit 1,000 rows, note that the tail was not assessed.
6. If no row exists, log "not listed, verified in the page source" with the date and set a recheck.
7. If a row exists but the source has no link, log that the row may be historical and send the source evidence to the seller.

No step assumes a waiting period, because the documentation gives none. Your team can set its own recheck dates as an internal policy.

A monthly limits note

A short note at the end of each monthly report spares you from re-explaining limits.

• Report scope: which table of the external links section.
• Row status: below 1,000 rows or at the cap.
• Export type: latest links, more sample links, or single table.
• Grouping note: how root domain grouping affected counts.
• Source of rel data: page source or tool.
• Verification: which pages were checked in the source, and when.
• Unknowns: what the report could not show.

Three decisions the limits can distort

Cleanup: acting on one odd anchor in a sampled table is acting on partial data. The Manual Actions help page says disavow is for links you cannot remove, so evidence and a removal attempt come first.

Confidence: a long list of linking sites does not show how many rows are stale, purchased or qualified. A clean-looking report does not replace the Manual Actions report.

Progress: a rising row count month over month reflects the sample and stale rows as much as new links. Measure progress with verified deliveries and the rate at which they disappear, and keep the report as supporting visibility.

Frequently Asked Questions

Why does my Links report stop at 1,000 rows?

The help page says tables are limited to 1,000 rows, so tables may be truncated on larger or more linked-to sites. Exports can reach 100,000 rows, but single-table pages export up to 1,000 rows.

Is the report a complete list of my links?

No. The help page says it is not a comprehensive list of every link, shows a sample, and may omit some URLs such as non-indexed pages or deduplicated URLs.

Does the report show nofollow?

No. The help page says the report does not specify whether a link is marked nofollow. Read the rel attribute in the page source.

If a link appears in the report, is it still live?

Not necessarily. The report includes links Google has found over time, and they may have since been removed. Open the page and confirm.

I bought two links from one publisher but see one row. Is one missing?

Not necessarily. Data is grouped by root domain and duplicate links are combined. Open the detail view for that site and look for each source page.

What can fill the gaps in the report?

The page source confirms existence, address, anchor and rel. Link tools add leads, the Manual Actions report shows manual actions, and your dated order records keep the delivery history.

Conclusion

The Links report is a sampled, capped, partly historical view with no rel data. Use it to find leads, not to issue verdicts. Confirm every link that matters in the page source, use link tools and the Manual Actions report for what the report cannot show, and add a short limits note to every monthly report so readers do not mistake a sample for the whole.

If you want links delivered with records you can verify, browse our backlink packages.