Link freshness and link age: do old links lose value?

Google records when a link first appeared, when it was last seen and whether it was deleted, so link age is tracked. No public source says old links fade simply because they are old. The larger risk is that links disappear, and that a profile stops gaining new ones.

In short

  • Google's leaked Content Warehouse documentation stores first seen and last seen dates and deletion flags for anchors, according to iPullRank's and Hobo's 2024 analyses.
  • A Google patent on historical data, reported as filed in 2003, describes scoring pages by when links appear and disappear and by how fast they accumulate.
  • In the Yandex source code leaked in January 2023, link age was reported as an explicit ranking factor, with older links treated as stronger.
  • No primary source states that a Google link loses value purely with age, and no source gives a decay rate.
  • An Ahrefs study dated February 2024 found that about two thirds of links seen over nine years were dead, so loss of links is a bigger practical problem than ageing.

Search engines track the age of a link, but nothing public says a link fades just because it is old. The evidence points to something more specific. Google records when a link first appeared, when it was last seen and whether it has been deleted. Analysts infer that links from fresh, well-maintained pages count for most, and that a link has to persist to keep counting. In Yandex’s leaked code, older links were reported as stronger, not weaker. So asking how old a link is gets you very little. The useful question is whether the link is still live, on a page that is still worth something, in a profile that is still growing.

People use the terms loosely, and they mean different things. Link age is how long a link has existed, so a link placed in 2018 and still live today is an old link. Link freshness covers two things. One is how recently the link was created. The other is how fresh the page carrying it is: whether that page is still updated, crawled often and held in the part of the index reserved for important documents. The second meaning is the one that gets lost in casual talk, and it is the one the evidence has most to say about.

Keeping the two apart matters because they can point in opposite directions. A link can be old and sit on a fresh page, such as a long-standing citation in a guide that is revised every year. A link can also be new and sit on a stale page, such as an insertion into an abandoned blog post. Those two cases are not the same, and the evidence treats them differently. If you only look at the date a link was created, the abandoned blog post looks like the fresher of the two, which is exactly the wrong way round once you consider the state of the page.

What the evidence says

Four sources bear on the question, and they are of very different kinds: a patent, a set of leaked field names, another search engine’s leaked code and a published study. The table keeps them apart so that the firmness of each one stays attached to what it says.

Source What it says How firm it is
Google patent on historical data, reported as filed in 2003 Scoring can use the dates links appear and disappear, the growth rate of links and changes in anchor text over time A patent. It shows the idea, not current use
2024 Content Warehouse leak Anchors carry first seen and last seen dates and deletion flags. A quality value ties an anchor to the index tier of the linking page. A separate structure tags links from news sites Documented fields. No weights
Yandex source code leak, January 2023 Link age listed as a factor, with older links stronger Reported by Search Engine Roundtable in 2023. Yandex is not Google
Ahrefs link rot study, February 2024 66.5% of links to sites in the previous nine years were dead Published figure from a named study

The leak analysis comes from iPullRank and Hobo, both published in 2024. The documentation lists data fields only. It does not prove a field is in use, and it contains no scoring formula. Our lesson on the leak and the court documents explains how to read it. The point to carry forward is that the first three rows describe what a search engine could do or what it stores, while only the last row is a measurement, and that measurement is about links disappearing, which says nothing about links growing weaker with time.

What is inferred, and labelled as such

From those fields, analysts draw three inferences. Treat them as reasoned readings, not as facts. The first is that the page matters more than the date. If an anchor’s quality bucket follows the index tier of the linking page, then a link on a page Google considers fresh and important counts for more than a link on a page that has sunk into a low tier. The age of the link itself is secondary. On this reading, two links created on the same day can end up with very different standing a few years later, depending on what happened to the pages around them.

The second inference is that links must persist. Google records when a link was last seen and whether it was deleted. A link that vanishes stops counting. This fits what practitioners see when placements are removed. The third is that news links are handled in their own way. A structure for anchors from news sources suggests that links from news sites are treated specially. Nothing shows how, so it would be a mistake to build a plan on the assumption that the special treatment is a bonus.

What the evidence does not support is a decay curve. You will see claims that a link loses a set share of its value each year. No source gives such a number for Google. A claim of that kind needs a formula or a measurement behind it, and the leak contains field names without a formula, the patent describes possibilities and the one hard figure on the table counts dead links. Anyone quoting an annual percentage loss has supplied it themselves.

On whether old links fade, the honest answer is that an old link can weaken for reasons that have nothing to do with its age. The article gets pushed deep into an archive and loses its internal links. The page stops being updated, loses its own traffic and is crawled less often. The site is sold, repurposed or filled with unrelated content. The page is deleted, redirected or set to noindex. Or the link itself is edited, removed or switched to nofollow. In each case the link’s age is incidental. The page changed. Time is simply what gives these things the chance to happen, which is why age and weakness get confused with each other.

The opposite case deserves equal attention. An old link on a page that is still read, still linked internally and still ranking is among the best links a site can have, and the Yandex evidence suggests at least one major search engine explicitly rewards that. Let us say a supplier of laboratory equipment was cited years ago in a university’s buying guide. If the guide is still revised and still sits in the department’s navigation, there is no evidence-based reason to think that citation has lost anything.

On new links, a steady flow is a sign that a site is still being talked about. The historical data patent is reported to describe link growth rate as a signal that can cut both ways: a surge may mean a page is newsworthy, or it may mean spam. That is the subject of our lesson on link velocity, and this page will not repeat it. For individual links, the working assumption is modest. A new link on a newly published, well-linked page starts life in a good position. A new link added to a dead page does not gain much from being new.

The Ahrefs link rot study, dated February 2024, found that 66.5% of links to sites in the previous nine years were dead. Whatever the true decay of a surviving link, most of the loss in a profile comes from links that no longer exist. That changes what you should spend time on. Arguing about whether a surviving link from some years ago is worth slightly less than it was is a debate with no data behind it, whereas the disappearance of links is measured, large and something you can act on.

The first job is to monitor. Check that links you earned or paid for are still live, still followed and on indexed pages. How to track backlinks sets out a schedule, and the backlink tools comparison covers the products that report lost links. The second job is to reclaim. When a link breaks because your own page moved, fix it with a redirect. When a publisher removed it, ask why. Link reclamation is usually the cheapest link work available, because the link was already won once and the remaining task is repair.

Two further habits follow from the same figure. For a paid placement, buy with a term: agree in writing how long the link stays. Insertions into old posts are removed more often than new articles, practitioners report, which is one of the trade-offs of niche edits. And keep earning. Because links are lost every year, a profile that gains nothing new shrinks, so standing still requires new links. Planning how many links you need should include replacing losses.

How to use this when you choose placements

These are working methods, not rules from Google, and they apply the reading that the page matters more than the date. Prefer pages that the host site still links to from its navigation, category pages or newer articles, since those internal links are the visible sign that the site itself still treats the page as current. Check that the host page is indexed and, for an existing article, that it still gets search traffic. A page that has dropped out of the index or draws no visits is the stale page described earlier, however recently your link was added to it.

Do not pay a premium for “aged” links as a product. The age of a link you did not earn naturally proves nothing about the page it sits on. For the same reason, do not strip or replace old links because they are old. If the page is healthy, leave it alone. Finally, record the date each link went live, so you can measure how many survive at 90 and 365 days. That record turns link loss from a vague worry into a number for your own profile, and it shows which sources and which kinds of placement hold.

How search engines evaluate links covers the other signals. For how other engines treat links, including Yandex, see links in other search engines. Link rot and related terms are in the glossary.

Common questions

Do backlinks lose value over time?

There is no primary evidence that a live link on a healthy page loses value just because it is old. Links lose value when the page is removed, deindexed, buried or abandoned, and many links disappear altogether.

Are new links better than old links?

Not as a rule. Analysts infer from the 2024 leak that links from fresh, frequently updated pages in Google's top index tiers count most. In Yandex's leaked code, older links were reported as stronger. Both age and freshness appear to be read as signals.

How long does a backlink last?

Many do not last. An Ahrefs link rot study dated February 2024 found that 66.5% of links to sites over the previous nine years were dead. Paid placements should come with a written term for that reason.

Does Google know when a link was created?

The 2024 leak documents date fields on anchors, including when a link was first and last seen and whether it was deleted. How those dates are used in ranking is not shown.

Do I need to keep building links once I rank?

Usually yes, at a modest pace. Links are lost every year, and competitors keep gaining them, so a profile that stops growing shrinks in relative terms.

Vendors to look at

  • BazoomEditor's pick

    Sponsored content and link marketplace, managed service

  • MotherlinkEditor's pick

    Backlink services, guest posts, niche edits, full SEO