Short answer

Backlinks aren’t permanent. Some survive a decade, others vanish in a month, and no one can give you a reliable survival rate because the industry doesn’t publish one. Links usually disappear for boring reasons: a site redesign drops old URLs, an article gets pruned or merged, or an editor changes the rules.

Ask a vendor how long their links last and watch what happens. You’ll get a guarantee period, which is a commercial term, or a percentage that appeared from nowhere. What you won’t get is a dataset. That gap is worth understanding before you plan a budget around links you’re assuming will still be there in three years.

How long do backlinks actually last?

There’s no honest single answer, and anyone quoting one precisely is guessing with confidence. What’s reliably true is that link profiles leak: a portion of every batch of placements will be gone within a year or two, and the rate depends far more on the type of host site than on anything you did.

The variables that matter are unglamorous. A link on a ten-year-old resource page maintained by a person who cares will outlive most companies. A link in a listicle on a content site that’s been sold twice has a much shorter expected life, because the next owner will consolidate posts and strip outbound links as a matter of routine. Editorial links you earned tend to persist longer than paid placements, mostly because nobody has a financial reason to revisit them — a trade-off worth weighing alongside the other ones covered in our honest assessment of buying links.

What we won’t tell you is a survival percentage. SerpInsight is new, we haven’t tracked placements over a multi-year window, and inventing a number like that would be exactly the behaviour this page is criticising.

What actually causes a link to disappear?

Site redesigns and migrations are the biggest single cause, followed by content pruning and article merges. Editor turnover, blanket nofollow policy changes and sites simply going offline account for most of the rest. Almost none of it is about you.

Roughly in order of how often we see each one:

  1. Redesigns and migrations

    A site rebuilds on a new platform, the URL structure changes, and old posts either 404 or get redirected to a category page. Your link doesn’t get removed so much as abandoned, and often the site owner has no idea it happened.

  2. Content pruning

    An SEO consultant tells the site that thin, low-traffic posts are suppressing performance. Two hundred old articles get deleted in an afternoon. Your placement was on one of them.

  3. Consolidation and merging

    Three overlapping posts become one canonical guide. The merged article keeps some outbound links and drops others, and the selection is rarely careful.

  4. Editor turnover

    The person who approved your insertion leaves. Their replacement inherits a site with paid links they didn’t agree to, and cleans house — sometimes politely, sometimes not.

  5. A site-wide policy shift

    The site gets nervous, or gets a manual action, and applies nofollow or sponsored attributes to every external commercial link at the template level. Your link is still there. It just stopped doing anything.

  6. The site goes away

    It’s sold, parked, expires, or the owner stops paying for hosting. Rare in any given quarter, permanent when it happens.

Removal cause Warning signs Preventable? What to do
Redesign or migration New branding, a “we’re refreshing” notice, a platform change visible in page source Partly — vet for site stability up front Ask the editor for the new URL; a 301 to the equivalent page usually preserves the link
Content pruning Post count dropping, older archives thinning, traffic decline on the host No, but predictable Prefer host pages that already rank and get traffic — those survive pruning
Article merge or consolidation Two similar posts on the host covering your topic Partly — avoid duplicate-topic hosts Ask for reinstatement in the merged article; editors often agree
Editor turnover New bylines, a changed contact address, an unanswered thread No Re-open the relationship rather than the argument; treat it as new outreach
Site-wide nofollow policy Every external commercial link on the site now carries rel=”nofollow” or “sponsored” No Claim the replacement; the link still sends referral traffic but stop counting it as equity
Site sold, parked or offline WHOIS changes, publishing stops, ads get heavier Partly — check for churn signals when vetting Nothing recoverable; this is what a replacement guarantee is for
Manual removal request An email asking for payment to keep the link live No Decline the renewal fee and take the replacement; paying twice sets a precedent

Why doesn’t anyone publish link decay data?

Because measuring it properly is expensive, and because the answer wouldn’t help anyone selling links. Tracking survival honestly means monitoring thousands of placements for years, publishing the failures alongside the wins, and controlling for host type — and the vendor with the data has every reason not to share it.

Think about what a real decay study would require. You’d need a large sample, several years of monitoring, a clean definition of “removed” that separates a deleted link from a nofollowed one from a 404’d host page, and the willingness to publish a number your sales page then has to live with. Tool vendors have the crawl data but not the placement records; link sellers have the placement records and no incentive.

The silence is convenient, and I’ll say plainly that it’s the one thing about this industry that irritates me most. It lets everyone imply permanence without ever claiming it. When you see a survival percentage on a sales page with no methodology attached, treat it as decoration rather than evidence, and ask how the sample was defined. The absence of an answer tells you what you need.

How long does Google take to notice a removed link?

Longer than most people assume. Google has to recrawl the host page to see that the link is gone, and recrawl frequency on an older article can stretch to weeks or months. The link keeps counting until that recrawl happens.

This cuts in both directions, and it’s the reason link data is always slightly out of date. A link you bought last week may not count for a month. A link removed last month may still be propping up a ranking today. Backlink tools have the same problem in reverse: they report a link as live until their own crawler revisits the page, which is why a tool showing 340 referring domains is really showing you 340 domains that had a link at some point in the recent past.

The practical consequence is that you can’t debug a ranking drop by checking whether links are live today. By the time you notice, the removal may have happened two months ago, or may not have been processed yet.

Why is cumulative attrition the real problem?

One lost link is noise. The problem is the compounding: if you’re building links at a steady pace and losing a slice of the older ones every year, a share of your budget is buying replacement rather than growth, and net referring domains can flatten while spend stays constant.

Picture it as a bucket with a slow leak. Building four links a month feels like adding forty-eight a year, but if attrition is taking a meaningful bite out of the previous years’ placements, your net position improves far more slowly than your invoices suggest. Sites that have been buying links for five years often discover their profile is much smaller than their spreadsheet.

Two things follow. First, older links are worth protecting — a placement from three years ago that’s been recrawled dozens of times is doing more work than a new one, so recovering it is usually cheaper than replacing it. Second, attrition should be in your model from the start. When you’re working out what a campaign costs, assume some of it evaporates; our breakdown of what backlinks actually cost is worth reading with that in mind.

How should you monitor your links for removals?

Quarterly is enough for most sites, monthly if you’re spending heavily. The workflow is: export your placement list, check each URL for the link and its rel attribute, flag the failures, and classify each one before you chase it. It’s tedious and it takes an hour or two a quarter.

  • Keep one source of truth: host URL, destination URL, exact anchor, placement date, price paid, editor contact.
  • Check the rendered page, not a cached tool report — tools lag by weeks and will show removed links as live.
  • Record the rel attribute each time, so a template-level switch to nofollow gets caught as a change rather than missed as a non-removal.
  • When a URL 404s, check whether the article moved before assuming it was deleted — a redesign often just renamed it.
  • Try recovery before replacement: a polite note about a merged article gets the link reinstated more often than you’d expect.
  • Set up Search Console and a backlink tool alert for lost referring domains as a backstop, not as the primary check.
  • Log every removal with its cause, so after a year you know which vendors and which host types leak.

That last one matters more than the rest combined. A year of classified removals tells you which supplier’s inventory is stable and which is churn dressed up as opportunity, which is information no vendor will hand you.

What should a replacement guarantee actually cover?

A guarantee is only meaningful if it names a period, defines what counts as a removal, and says who does the checking. Ours is twelve months, free replacement, with SerpInsight doing the monitoring rather than waiting for you to notice — but the terms matter more than the length.

Questions to put to any vendor, including us:

  • Does a link switched to nofollow count as a removal, or only a deleted one?
  • Who monitors — you, or them? A guarantee you have to police yourself is a refund policy with extra steps.
  • Must the replacement match the original tier’s DR and traffic floor, or can it be anything they have spare?
  • How long do they have to deliver the replacement once a removal is confirmed?
  • Is the host site excluded if it goes offline entirely, which is when you most need the cover?
  • Is a refund available instead of a replacement if their remaining inventory doesn’t suit you?

The vetting standard is what does most of the work here, and a guarantee is the backstop rather than the plan. Requiring the host page to be indexed and ranking, insisting on 1,000 monthly organic visits minimum, and checking that the site has real topical depth all select for pages nobody’s going to prune — which is roughly the reasoning behind how we approach placements. If you’re assessing hosts yourself, our guide to judging backlink quality covers the same checks in more detail.

Key takeaways

  • No backlink is permanent, and no one can give you a credible survival rate because the industry doesn’t publish decay data.
  • Redesigns and migrations remove more links than anything else, followed by content pruning and article merges — very little of it is personal.
  • A link switched to nofollow at the template level is a removal in every way that matters, and most guarantees are vague about it.
  • Google can take weeks to notice a removal because it has to recrawl the host page first, so live-link checks always lag reality.
  • Cumulative attrition, not any single lost link, is what quietly turns a growth budget into a replacement budget.
  • Monitor quarterly, classify every removal by cause, and try recovery before replacement — reinstatement requests work more often than expected.

Common questions about link decay

Is a link that turns nofollow the same as a lost link?

For ranking purposes, treat it as lost — Google describes nofollow as a hint it may ignore, which is not something to build a forecast on. It still sends referral traffic, so a nofollowed link on a page with real readers isn’t worthless, just worth something different from what you bought.

Should I pay a site to keep a link live?

Generally no. Once you’ve paid a renewal fee on one placement you’ve taught that site your links have a recurring price, and the second invoice tends to arrive faster than the first. The exception is a genuinely irreplaceable host page that drives referral traffic you can measure.

How often should I audit my backlink profile?

Quarterly for most sites; monthly if you’re placing more than about ten links a month, because the list gets unmanageable otherwise. An audit of forty placements takes roughly an hour if your tracking sheet is in order.

Do older links carry more weight than new ones?

Not because of age itself, but an older link has usually been recrawled many times and sits on a page that’s proven it survives redesigns and pruning. That stability is worth protecting, which is why chasing reinstatement of a three-year-old placement beats buying a fresh one.

Does a 301 redirect on the host page preserve my link?

Usually yes, if the old article redirects to a genuinely equivalent page that still contains your link. A redirect to the homepage or a category listing is effectively a removal, because the link itself no longer exists anywhere.