If you linked to a site years ago and it has since changed owner and started hosting junk, Google does not penalize you for it. That was John Mueller's answer to this specific question in a Reddit discussion. The original question has since been removed, but the discussion preserves the context. This is about an old destination changing, not permission to build manipulative links.
What was actually asked
The question is the one every older site eventually runs into. You publish an article, you cite a useful source, and years later that domain expires, gets bought, and turns into a spam page or something worse. Your link is still there, still pointing at it.
Google's position is that this does not count against you.
Why that is the expected answer
Routine link decay and deliberate manipulation are different problems. Google's spam policies still apply to links created primarily to manipulate rankings. Mueller's answer addresses a previously useful destination changing after publication.
The alternative would be unworkable. Every publisher would be responsible for the future behavior of every domain they ever cited, forever, including domains they have no control over and no way to monitor.
We ran the check on our own archive
Advice about link maintenance is easy to give and rarely tested, so we tested it here. In September 2026 we pulled every outbound link out of this site and requested all of them.
The archive at that point held 26 articles carrying 70 unique external links across 46 domains. Here is everything that came back.
| Result | Links |
|---|---|
| 200, loaded normally | 64 |
| 403, refused an automated client | 5 |
| No response at all | 1 |
| Redirected, then loaded | 3 |
| Confirmed dead (404 or gone) | 0 |
Zero. Not one link in the archive is actually broken.
That result is less impressive than it sounds, and the reason matters more than the number. This archive is weeks old. Link rot is a function of time, and almost nothing we cited has had the chance to expire yet. The interesting part is not the zero. It is the six links that a checker reported as problems.
What an automated checker gets wrong
Six of seventy came back looking broken. None of them are.
Five returned 403. That is a server refusing the request, not a page that stopped existing. Large sites increasingly block anything that does not look like a browser, and a link checker never does. We retried the OpenAI page from their announcement of ads in ChatGPT with a full desktop browser user agent and it still returned 403, while the same URL opens normally in an actual browser. The block is on the client, not the content.
One returned nothing at all. A post on the Adobe business blog timed out with no status code. We checked further before writing it off: DNS resolves, the domain root answers 200, and only that one path refuses to respond. The article is there. The edge in front of it will not talk to us.
So an automated pass over this archive produces a list of six URLs to fix, and acting on that list would have deleted six working citations. The checker is useful for narrowing down where to look. It is not useful as a verdict.
This is worth knowing before you run one over a large archive, because at scale nobody opens all the flagged URLs by hand, and the temptation is to trust the report.
How to run it on your own site
There is nothing to install. Extract the links, request them, and read the codes yourself:
- Pull every outbound URL out of your content files or your sitemap.
- Request each one following redirects, with a timeout, and record the status.
- Sort the results by code rather than by pass and fail.
- Open every non-200 in a real browser before touching anything.
That last step is the whole method. Everything before it is just narrowing the list.
Read the codes for what they mean. A 404 or 410 is a genuinely missing page and needs a decision. A 301 is fine, though updating the link to its destination saves the reader a hop. A 403 or a timeout usually means you were blocked, and the only way to know is to look. A page that returns 200 can still be dead in the sense that matters, because a parked domain full of ads answers 200 all day.
What to do anyway
Nothing, if you are only worried about rankings. But there is a reason to check your outbound links that has nothing to do with Google.
A reader who follows a citation and lands on a parked domain full of ads learns something about your article, and it is not good. The link was your evidence. If it now leads nowhere, the claim it supported reads as unverifiable.
So the practical version is short:
- Check the outbound links in your most-read articles once or twice a year. Not all of them, just the ones people actually open.
- Replace a dead source with an archived copy, or with a different source that still stands.
- If a link now points at something harmful, remove it. For the reader, not for the algorithm.
- Keep using
rel="sponsored"on paid or affiliate links andrel="nofollow"where you cannot vouch for the destination. That is about disclosure, and it has not changed.
When the numbers start to change
An archive of a few dozen articles cited over a few weeks will read close to clean, as ours did. The picture changes with age, and it changes unevenly.
Links to documentation and policy pages move constantly, because the companies publishing them reorganise their own sites. Links to news articles hold up better than links to company blogs, because publishers keep archives and companies quietly retire posts. Links to small independent sites are the ones that vanish outright, since a lapsed renewal is all it takes.
The practical consequence is that the first real audit of an archive is worth doing at around two years, not at three months, and it is worth doing on the handful of pages that carry traffic rather than on everything.
An honest illustration sits in this very article. The Reddit thread quoted at the top had its original question removed after we cited it. The URL still answers 200 and the discussion is still readable, so no checker would flag it, and the citation still works because the surrounding replies survived. That is the shape most decay takes: not a dead link, just a source that quietly became weaker than it was on the day you used it.
The part worth remembering
Link rot is not a penalty risk. It is a credibility risk, and it grows quietly as an archive gets older. The fix costs an hour a year and only matters on the handful of pages that carry your traffic.
The trap is not missing the broken links. It is trusting a tool that calls six healthy ones broken.
Related reading
For the distinction between routine maintenance and manipulation, see our analysis of Google spam updates and iGaming SEO.
