Backlinks decay naturally — pages get deleted, redesigns drop links, editors update content. Each lost link removes whatever ranking signal it was passing — though on any single link that is usually far too small to observe, and normal decay is not something to panic about. Recovery is faster and cheaper than building new links: the publisher already linked once. The workflow is monitoring, root-cause classification, and targeted outreach. Part of the Backlinks audit programme.
Sources: - Ahrefs / Semrush / Majestic "Lost backlinks" reports - GSC → Links → compare month-over-month - Your aiwebpageseo backlinks watch (automated alerts) Weekly review: - Newly lost links since last check - Total backlink delta - High-value losses (DA 50+) need urgent action
Cause Action
─────────────────────────────────────────────────────────────
Linking page 404 Find new equivalent page, request relink
Page exists, link removed Outreach to publisher
Link rewritten to competitor Critical — competitive intelligence
Site no longer exists Lost permanently, archive only
Link nofollowed Nofollow became a hint rather than a directive in 2019,
so Google may or may not still use it. Low priority.
Anchor changed to brand Usually fine, less authority impact
Site redesign / migration Outreach during transition window
Pattern that works: - Identify the publisher contact (author, editor) - Reference the specific old page where link existed - Explain the link was helpful for their readers (why) - Note that the link now goes to (404 / removed) - Offer the current correct URL - Acknowledge their time Success rate: 30-50% on legitimate publishers. Higher when you reference specific reader value. Lower for sites that auto-purge old pages.
Don't chase every lost link. Prioritise: - DA 50+ → outreach within 1 week - DA 30-50 → outreach within 2-4 weeks - DA < 30 → batch monthly outreach, skip if no contact - Toxic / spammy → don't try to recover; let them stay gone
Some causes are preventable:
Per month: - Lost links detected - Outreach attempts - Restorations confirmed - Restoration rate Healthy: 30-50% restoration on actively-outreach'd losses. Below 20%: outreach quality low, or targets are auto-purgers. Above 50%: well-built relationships and good outreach.
The classification table treats loss as something that happens to you. Run the analysis honestly on most sites and a striking share of the losses turn out to be self-inflicted.
The publisher did nothing wrong. You moved the target. - A redesign changed your URL structure - Old URLs 404 instead of redirecting - A redirect chain grew past what Google follows - Somebody "tidied up" old redirects after a migration - A page was deleted rather than redirected to its successor
Every one of those kills external links to the page without the publisher touching anything. And every one of them is preventable, at essentially no cost, by one rule: keep redirects indefinitely for any URL that ever had an external link.
Redirects are cheap to serve and expensive to remove. The day you delete them, every backlink pointing at the old URL starts returning a 404 — and you have discarded the value of a migration you already paid for, in the name of housekeeping nobody asked for.
A weekly lost-links report will hand you a list, and working the list top to bottom is a good way to spend a quarter recovering things that never mattered.
Links decay for ordinary reasons. Pages get deleted, sites get redesigned, editors rewrite articles, publishers go out of business. A steady trickle of loss is simply what a real link profile looks like — and the value of any individual link is usually far too small to observe in your rankings either way.
Worth acting on:
- A link from a site that sends you actual referral
traffic. Check analytics: this is the clearest
signal of a link that was doing something.
- A link lost because of a fixable technical cause
(your 404, your redirect, their migration).
- A link that was rewritten to point at a competitor.
That is not decay — that is an editor changing their
mind, and it is worth understanding why.
Not worth chasing:
- The site no longer exists.
- A link you cannot remember earning, from a site
that never sent you a visitor.
- Ordinary editorial churn on a page that was rewritten
for reasons that have nothing to do with you.
The referral-traffic test is the most useful filter here, and almost nobody applies it. A link that has sent you real people is a link with demonstrable value, whatever a third-party authority score says about it. A link that has sent you nobody in three years is a number in a tool.
Recovery has a far better response rate than cold link-building, and it is worth understanding why, because the reason dictates the email you should write.
You are not asking for something. Somebody already made the editorial decision that your page was worth citing — and then a migration, a redesign or a rewrite broke it by accident. Pointing that out is a favour. It is closer to reporting a bug than making a request, and it should read that way.
What works:
- Be specific: this page, this link, it used to point here
- Be useful: here is the URL that replaced it
- Be brief: three sentences, no pitch, no attachment
- Be human: to the author or editor, not to a form
What fails:
- A template that says "I noticed a broken link on
your site" and then pitches four other pages
- Any email that reads like link-building, because
editors receive twenty a day and delete all of them
And when the link genuinely cannot be restored — the site is gone, the page is gone, the editor has moved on — let it go. The competitive intelligence in the report is worth more than the last twenty percent of recoveries: a link rewritten to point at a competitor is telling you something about their content that no keyword tool will.
Lost-link reports are full of authority language — DA, DR, authority lost, equity leaked — and it is worth separating the parts that describe Google from the parts that describe a vendor.
Real:
- Links pass ranking signals. PageRank is a real,
published model. A link that existed and now does
not is a signal that existed and now does not.
Vendor metrics:
- DA, DR, TF, CF and their cousins are third-party
scores. No search engine publishes, endorses or
uses them. A "DA 50" link is a statement about
a tool's model, not about Google's opinion.
Folklore:
- Percentages of authority lost per event
- Precise recovery timelines
- The idea that nofollow "leaks" anything —
nofollow became a hint in 2019, and what Google
does with those links it has not said
None of this means ignore the scores. They are a rough sort order, and a rough sort order is genuinely useful when a report hands you four hundred rows. It means do not build a business case on them, and do not spend a fortnight recovering a link because a number next to it was large.
The check that beats every score: did this link ever send you a person? Referral traffic is measurable, it is yours, and it is a fact rather than a model. A link that has sent real visitors is demonstrably doing something. A link with an impressive score that has never sent you anybody is, at best, an unverified claim.
Weekly review is right, and it is worth setting the expectation that most weeks the answer is “nothing to do” — because a monitoring habit that always produces a task list will always produce busywork.
Normal decay needs no response. The alerts worth building are narrow:
Everything else can wait for a quarterly look. Link recovery is real work with a real return, and it is also the kind of work that expands to fill whatever time you give it — which is time not spent publishing the thing that would have earned the next link on its own.
2-5% of total backlinks per month for mature sites. Higher if you've had recent content changes or site migrations. Steady losses are normal; spikes need investigation.
No — losing toxic links is good. Some natural loss is fine. Prioritise DA 30+ recoveries; ignore low-quality losses.
Once restored, Google may take 2-8 weeks to re-crawl and update authority. Effect on ranking similar timeline.
Yes — you're requesting correction of a previous link, not buying new placement. Most editors view it as housekeeping help.
Some of it, and less than the reports imply. Links decay for ordinary reasons — pages get deleted, sites are redesigned, editors rewrite articles, publishers go out of business — and a steady trickle of loss is what a normal link profile looks like. The losses worth acting on are the individually significant ones: a link from a site that genuinely sends traffic or carries real weight, lost for a reason you can fix. Chasing every lost link from every small site is a way of spending a week to recover something you could not have measured in the first place.
Because the hard part has already happened. Somebody decided your page was worth citing, made the editorial judgement, and published the link. Getting that decision reversed by accident — a migration, a redesign, a rewrite — is not the same as being told no. So the outreach is a correction rather than a request, and the response rate is far better than cold outreach: you are pointing out that a link they chose to include has broken, which is a favour rather than an ask.
Your own migrations, more often than teams expect. A redesign changes URLs, the old ones 404, and every external link pointing at them dies — not because the publisher removed anything, but because you moved the target. That is entirely preventable, and it is why redirects should be kept indefinitely for any URL that ever had an external link. Deleting old redirects to tidy up is one of the few genuinely self-inflicted SEO wounds, and it discards the value of migrations you already paid for.
aiwebpageseo.com is a data-driven SEO and AEO (Answer Engine Optimisation) platform providing a free suite of technical website tools. Rather than relying on AI-theorised assumptions, the platform analyses live URL performance, delivering objective diagnostics, page speed metrics, CLS debugging, and site crawl data alongside actionable technical tutorials.