Free tool
Enter the page where your backlink should live and the target URL it should point to. The check reports HTTP status, link presence, anchor text, rel attribute, and indexability.
Definition
A live backlink is one that is present in the HTML a crawler receives, on a source page that returns a 200 status, pointing at the target URL you agreed, with the rel attribute you expected, on a page that is indexable and canonicals to itself. Miss any one of those conditions and the link either does not exist or does not do what you think it does. This is why link count on its own is a poor metric: a report saying 120 backlinks is only meaningful if somebody verified those six conditions recently, and usually nobody has.
01
Present, not just visible
The link must be in the served HTML. One injected by JavaScript after load may not be counted reliably.
02
On a page that resolves
A source page returning 404, 410, or a redirect chain takes your placement with it.
03
Indexable
A noindex directive means the page carries no ranking signal. Your link is there and it counts for nothing.
04
Canonically self-referencing
A page canonicaling elsewhere has delegated its signals. The link looks fine and behaves differently.
The eight checks
These are the checks worth running, in rough order of how often each one catches something. If you build your own monitoring script, this is the specification.
Start with the HTTP status of the source page. A 200 means the page is served normally. A 404 or 410 means the placement is gone along with the page. A 301 or 302 needs following, because a redirect to a relevant new URL that still contains your link is fine, while a redirect to a homepage or a category page usually means your placement was collateral damage in a migration.
A placement that switched to nofollow is the most common form of value loss that looks like nothing happened.
Second, link presence. Does your target URL actually appear in the returned HTML. This is the check that catches deliberate removals and quiet editorial cleanups, and it is the one most people mean when they ask whether a link is still live.
Third, anchor text. Compare what is published against what you agreed. Drift happens more than you would think, particularly with insertions, where an editor rewrites the sentence around your link and the anchor becomes click here or the publisher's own phrasing. Not fatal, worth knowing, and much easier to get corrected in the first month than the sixth.
Fourth, the rel attribute. Dofollow, nofollow, sponsored, or ugc. A placement that silently switched to nofollow is the most common form of value loss that looks like nothing happened, because the link is still there and a casual glance finds no problem at all.
Fifth, page level noindex. If the source page carries a noindex meta tag or header, the page is excluded from the index and your link passes nothing. Sites do this to thin content, tag archives, and old campaign pages, sometimes years after publishing.
Sixth, the canonical URL. If the page canonicals to a different URL, its signals consolidate there. Worth checking whether your link also exists on the canonical target, because sometimes it does and you are fine, and sometimes it does not.
Seventh, does your target URL still resolve on your own side. This one is embarrassing when it fails, and it fails regularly: you restructured your site, the page your partner linked to now redirects twice or 404s, and you have wasted a placement through your own housekeeping.
Eighth, log the timestamp. Not a check exactly, but the thing that turns checks into a record. Without a last checked timestamp you cannot tell the difference between a link that passed yesterday and one nobody has looked at since spring, and you cannot report survival rate at all.
Why manual checking fails
Checking one link by hand is easy. Open the page, use find on page, view source if you want the rel attribute, done in ninety seconds. The reason manual verification breaks has nothing to do with difficulty and everything to do with arithmetic.
At 90 seconds per link, 100 placements is two and a half hours. Doing that weekly is ten hours a month of unbillable work that produces nothing new most weeks, which is precisely why it gets skipped. And a check you skipped is indistinguishable from a check that passed, so your inventory slowly becomes a set of assumptions.
There is also a subtler problem, and it is the case for running the checks on a schedule: humans check the wrong links. Left to instinct, you check recent placements, because they are top of mind, and you rarely revisit the ones from a year ago. But survival curves do not work that way. Links mostly die well after publication, during redesigns, content audits, and editor changes, so the placements least likely to be checked are the ones most likely to be dead.
The third failure is silent state change. Nobody emails you when a link goes nofollow or a page gets a noindex tag. There is no notification for a canonical change. These are exactly the changes a scheduled check catches trivially and a human eyeballing a page misses completely, because the page looks identical.
So the honest recommendation is: use a one at a time checker for spot checks and disputes, which is genuinely what it is good for. Automate anything above roughly 50 links, whether with a script you write yourself or a tool. The checks are simple enough to build, and the hard part is scheduling, storage, and surfacing changes rather than the fetch itself.
When a spot check is the right tool
Diagnosis
The response depends entirely on which check failed, and the wrong response wastes an outreach opportunity you only get once.
The page is gone. Check whether the content moved to a new URL, because if it did, your ask is simple and reasonable: please carry the link over to the replacement. If the content is genuinely retired, this is a loss to record rather than a recovery to chase.
Deliberate or accidental removal. This is the highest value recovery case and the most time sensitive, so start from the lost backlink recovery email. Contact the editor within two weeks, be polite, offer them an easy out in case removal was intentional, and offer to supply a replacement sentence to make saying yes trivial.
Usually a sitewide policy change rather than anything about you, which means asking for an exception rarely works. Record it, decide whether the placement is still worth its slot in your reporting, and factor the publisher's new policy into whether you trade with them again.
Low urgency, easy fix, and worth asking early. An editor who just published will usually adjust an anchor without complaint. One who published nine months ago will wonder why it matters now.
The placement passes nothing. Worth a polite question, since noindex on an article is often unintentional and the result of a plugin or template change rather than a decision. Sometimes you are the one who tells them their content is not being indexed.
Check whether your link also appears on the canonical target. If it does, you are fine. If not, ask whether the link can be added to the version they consider primary.
Fix your side first, then confirm. Redirecting the old target to the right page usually preserves most of the value, and it is faster than asking a partner to edit anything.
Failure modes
None of these send you a notification. That is the whole problem.
An editor cleans up outbound links in an old post, or a new editor arrives with different opinions about linking out.
The article's URL changes or the archive gets pruned. Your placement disappears with the page rather than by decision.
The sentence around your link gets edited and the anchor becomes generic. The link survives, the relevance signal weakens.
A sitewide nofollow or sponsored policy lands. The placement looks untouched and stops passing authority.
The page gets excluded from the index or delegates its signals elsewhere. Invisible unless you check the head of the document.
A partner removes their link because the one you promised never arrived. The most preventable cause on this list.
Terminology
Plain definitions for the fields, since several of them get conflated in backlink tool interfaces.
Questions
An extension inspects the page you are currently looking at, in your browser, with your cookies and your ad blocker in play. A server side check fetches the page the way a crawler does, which is what you want, because the version a search engine sees is the version that counts. Extensions are also useless for checking a hundred links.
HTTP status of the source page, whether your target URL appears in the returned HTML, the anchor text wrapping it, the rel attribute, any page level noindex directive, and the canonical URL. Those six fields answer nearly every question worth asking about a single placement.
Because a page that canonicals to a different URL is telling search engines to consolidate its signals elsewhere. Your link technically exists, on a page that has effectively delegated its authority somewhere else. This one catches people out, since the link looks perfectly fine when you visit the page.
Most often because the link is injected by JavaScript after the page loads, and a plain HTML fetch never sees it. Sometimes because the site serves different markup to non-browser user agents. Either way it is worth knowing, because a link that only exists after JavaScript execution is a link some crawlers may not count reliably.
No, but be clear about what you are getting. A nofollow link still sends referral traffic, still builds brand awareness, and still appears in a reader's path. What it does not reliably do is pass ranking authority. The reason to check the rel attribute is that placements switch from dofollow to nofollow without anyone telling you, usually after a publisher policy change.
Weekly for anything you care about, daily for placements you would fight to keep. The instinct is to check new links often and old links never, which is backwards: placements are most likely to die in month four or later, during a redesign or a content cleanup, not in week one.
Not with this tool, which handles one link at a time. The bulk domain quality checker covers a list of domains, and continuous scheduled monitoring across every link you have built is the paid product. If you have more than about 50 placements, a one at a time checker stops being practical fairly fast.
Related
For a list rather than a single link, start with the bulk checker or the monitoring checklist.
Bulk domain quality checker
Run mechanical checks across a list of domains at once.
Read moreMonitoring checklist
Free checklist covering all eight checks described above.
Read moreBacklink monitoring
How scheduled checks work, and cadence by plan.
Read moreLost link recovery
The recovery queue and outreach that gets placements restored.
Read moreRecovery email template
Free email template for asking an editor to restore a link.
Read moreBalance calculator
Check whether an exchange partner has reciprocated fairly.
Read moreFurther reading
What each rel value means, and what to do once a check comes back wrong.
Link attributes
Dofollow is not an HTML attribute, and the three that are real describe a relationship rather than a quality. What each value means, which to use when, and the part nobody mentions: attributes change after you earn the link.
12 min read · Updated Aug 2026
Backlink placement
Thirty placement labels collapse into two questions and four quadrants: is the link in the main content or the template, and did a person put it there or a system. Plus the sitewide arithmetic that inflates every backlink report.
15 min read · Updated Aug 2026
Lost Backlink Recovery
A four step playbook to triage removed backlinks, diagnose the cause, and recover link equity while the window is still open.
9 min read · Updated Aug 2026
Backlink types
The type tells you how a link was acquired, not whether it is any good, because every category spans the full quality range. Here are 25 of them grouped by acquisition, plus the column nobody publishes: how each one eventually disappears.
18 min read · Updated Aug 2026
LinkSwapy runs all eight checks on every placement your team has built, on a schedule, and tells you the moment one changes.