Expired domain backlinks decay before the auction closes

Expired domain backlinks are the inbound links a domain picked up before its registration lapsed — links that still point at the domain name, and that whoever registers it next acquires along with it. None of them move. They stay on pages other people own.

A listing sells that record, priced on a score a crawler worked out while the site was still up. By the time the name can be yours, policy has required it to resolve to nothing for weeks. One thing worth having before you bid: the price on the listing is the smallest number in this transaction — redirect the name at a site you already own and that site carries the consequence, while the domain is the part you can abandon tomorrow. What follows: what survives that gap, what Google's expired domain abuse policy covers — narrower than nearly every page here reports — and six checks worth doing before you bid.

An expired domain is a link profile in mid-decay

Two things change hands at a domain auction: the registration and the name. That's the list.

The backlinks don't transfer, because they were never the previous owner's property either. Each is a string of HTML inside a page belonging to somebody else — a university department, a trade magazine, a hobbyist who last logged in around 2016. The old owner had no say over any of them. Neither will you. What you acquire is the historical fact that those pages once typed that name.

Which is why authority is the wrong noun. It sounds like a stored quantity, something in a tank that comes with the keys. What's there is a scattered set of pages under separate control, each editable or unpublishable on its own schedule, none of them told the domain changed hands.

They also decay faster once the domain goes down, which a listing has no way to show. An editor running a broken-link sweep across their archive finds an outbound link that no longer resolves and does the responsible thing — pulls it, or swaps it for a live source. The expiry is what puts your future backlinks in front of them. Every month the name sits dead, some share of the profile is tidied away by people acting in good faith.

That's an accelerant on top of ordinary attrition, and ordinary attrition is already severe. Ahrefs sampled 2,062,173 websites and found 66.5% of the links pointing at them, back to January 2013, had rotted — the largest cause, 47.7%, being the linking page dropping out rather than the link being edited out. That study is stamped February 2024 and its data stops earlier still, so read it as the shape of the web, not a measurement of today's.

How long is the domain dark before you own it?

Long enough to matter, and it isn't guesswork: the governing documents require it.

Start before the deletion, the part that catches anyone buying at a registrar's expiry auction and assuming they're getting a running site. Under ICANN's Expired Registration Recovery Policy, a registrar holding an expired name for eight days or more must cut it off: "For at least the last eight consecutive days (after expiration) that the registration is renewable… the existing DNS resolution path… must be interrupted by the registrar…". Put a holding page there instead and it "must conspicuously indicate that the domain name registration is expired". The site is required to be off, and to say so, through the last eight days of the window expiry auctions run in.

If nobody renews, the registrar deletes the name and it enters REDEMPTIONPERIOD. The same policy sets 30 days for every gTLD registry and requires that "the registry must disable DNS resolution". Verisign's functional specification for .com says the same in the registry's own terms — such a name "will not be included in the zone file" — and adds five calendar days of PENDING DELETE, also out of the zone file, before it's purged and released first come, first served.

Our arithmetic on those two documents, not a figure either prints: a .com running the full course has no site on it for at least 43 consecutive days before it reaches the drop — eight of them served by the registrar as an expiry notice, then thirty-five with no DNS answer at all, out of the zone file by the registry's own terms. Every referring page pointed at a dead destination for the whole stretch, and every crawler that called recorded that. Bid at the expiry auction and the first eight already happened; catch it at the drop and all forty-three did.

Narrower than it looks in one respect: the 30 days binds every gTLD registry, the five-day pending delete is a .com value, and country-code registries write their own rules. What doesn't vary is the direction. Nobody can tell you what share of a profile breaks in the window — some of it does, every time, and no listing accounts for it.

Why the DR on the listing is the wrong number

We sell monitoring. There's no domain on this site to buy and no link index behind this page — worth saying before the next three paragraphs, not after them.

The least arguable version comes from Google. In the March 2024 post that introduced the expired domain policy, answering a question about site-level signals, they wrote that "some third-party services provide 'reputation' or 'authority' scores for sites. These don't correspond to any of Google's own signals nor come from Google."

So the figure is a vendor's estimate, and estimates are useful. The problem is when it was made. The score comes out of a crawler's index, and an index is a photograph rather than a window: it holds links nobody has been back to look at, pages that 404 now, links the host quietly qualified with nofollow while it was still alive. Recrawl budget goes to pages that change and sites that matter, and a dead domain sits at the bottom of that queue. The number isn't wrong. It's late, and late in one direction only.

Which also means the figure can fall without anything happening to the domain. Next time that vendor refreshes its index, dead links get recounted as dead and the number you paid for drops. The asset didn't change. The measurement caught up.

Referring-domain counts fail differently. A large share of them on any domain are machine-made — whois mirrors, statistics clones, appraisal calculators, scanner reports. An expired domain collects one category the rest of us don't: the "recently dropped domains" listings, which link to it precisely because it expired. The count rises as the asset gets worse. We've covered what DA and DR actually measure and what a scoring tool means by toxic elsewhere.

Google named this one, and named it narrowly

Google's spam policies, read on 11 September 2026 on a page stamped last updated 28 August, carry a section headed expired domain abuse. "Expired domain abuse is where an expired domain name is purchased and repurposed primarily to manipulate search rankings by hosting content that provides little to no value to users." Its examples: affiliate content on a site previously used by a government agency, commercial medical products on a site previously used by a non-profit medical charity, casino content on a former elementary school site.

Read the verb. Hosting. That section is about what you publish on the domain once it's yours — not the purchase, not the price, not the links pointing at it.

Its position says the same. Expired domain abuse is a top-level section, filed between doorway abuse and hacked content, and it arrived on 5 March 2024 alongside scaled content abuse and site reputation abuse. Three content policies; link spam wasn't among them. And the link spam section — the one everyone cites on this topic — mentions none of it. Not "expired", not "dropped", not "aged". No domain history, no redirects. Absent from the definition and from all eleven examples. I read it term by term, because an absence is only worth reporting if you checked.

Two more things the announcement settles. Several pages date this policy to May 2024, but the sentence "this new policy will take effect starting May 5, 2024" appears once in that post, inside the site reputation abuse section, about site reputation abuse. Expired domain abuse was enforceable the day it was published. And the exemption is plain: "It's fine to use an old domain name for a new, original site that's designed to serve people first."

Everything on the first screen for these searches is published by a registrar, a host, a marketplace, a drop-catcher or a finder tool. That's who writes about domains. It goes some way to explaining how one narrow content policy keeps getting reported as a rule about links.

Three things people do with an expired domain, and three different rulebooks

"Is it risky?" has no answer until you say what you intend to do with it. The three common plays sit under different rules and put different things at risk.

Rebuild the site and link out from it. Publish on the domain, point links from it at something you own. This is the play the policy is aimed at, and it's link spam as well if those links exist to move rankings — the only one of the three that collects both. Done properly it isn't a link tactic but a months-long editorial project, on a domain that may simply decline to re-index. The network version is PBN backlinks.

Redirect it at your money site. Five minutes of work, which is why it's the common one. Named in neither policy, judged on relevance rather than policy — the next section. The exposure doesn't sit on the cheap domain. It sits on the site you pointed it at. If the money is a client's, that's also the sentence you'll be explaining in the monthly report.

The third doesn't feel like a play at all, which is why so many people are in it: you buy a placement on somebody else's rebuild. If you've paid for a link on a site that turned out to be a repurposed expired domain, the policy isn't your question — you're a link buyer, and that argument has its own article. What helps here is reading the seller rather than the asset, and we keep a checklist for it.

What a 301 from an expired domain is judged on

Somewhere between 90% and 99% of link equity passes through a 301. You've read that. The figure is quoted everywhere and sourced nowhere. Google's redirect documentation, last updated in April 2026, says nothing about equity, authority or PageRank through a redirect — no percentage, no mechanism, no direction. Google's own people have been read both ways on it over the years, and neither reading was ever written down. What the documentation doesn't contain is the whole finding: nobody quoting 90% is quoting Google.

And you get no reading either way. A redirect that passes nothing looks exactly like one that passes a little: no error, no message, no line in any report. Whatever you conclude from the traffic afterwards, you'll be concluding it from something else.

Drop the percentage and a better question surfaces. Does the old domain's subject have anything to do with where you're sending it? A defunct regional newspaper redirected at a regional news site is one thing; the same newspaper redirected at a casino is another, and no configuration turns the second into the first. Relevance has no settings page. The cost is lopsided too — you can abandon the domain tomorrow, while the site you pointed it at carries your revenue.

There's a fourth party in all this that nobody writes for, and it's the most defensible of the lot: the company buying back a name it used to own, or one a squatter took after a rebrand. It has the cleanest version of this job and usually does it wrong. Blanket-redirecting an old domain to a new homepage throws away the only precision available: a link to /2019/annual-report sent to a homepage is now a link to a homepage. Map URL to URL for everything that had inbound links, then trace the chain end to end — old domains often carry redirects left by a previous owner, so you're adding a hop rather than starting clean. And if what you've inherited is a genuine mess, disavowing is a narrower instrument than it looks.

Six checks, and what each one kills

None of this needs a subscription and all of it fits in an evening. Each one is a disqualifier rather than a score: passing all six doesn't say a domain is good, only that this one isn't already ruined.

What passing all six looks like: a defunct company in your own niche, trade-press links still live on pages still being published, and a page on your site that replaces what they sold. People make that decision and it's an ordinary one. It just isn't what most listings are selling.

Three things the table can't hold. Insist on live counts rather than index counts; a marketplace quoting 400-odd referring domains is quoting the stale index the price came from. Follow where each link actually arrives, because one arriving through a hop somebody else controls can be repointed the week after you buy. And a site left unpatched for years before it expired has often been compromised and used to host injected outbound links — visible in the Wayback record, but in the footer of a late capture, not the article body.

DomCop published a walk-through of screening 69 expired domains under $100. The page shows no date, though its own metadata puts it at 11 June 2026; we read it on 11 September 2026. It reviews ten of the sixty-nine by hand and shortlists three of those ten, so take it as a list of disqualifier categories rather than a survival rate. Those: a casino redirect in recent history, spam ad links early on, a name tied too tightly to a well-known company, an unexplained language shift, and one bound so closely to a government program that nothing else could be built there.

What we can't read from outside

LinkGuard watches links you already know about. You give us the donor URL, the target, the anchor and the rel you were promised; we re-check them on a schedule and tell you when one of those stops being true. A domain you're thinking of buying is the mirror image: links you don't know about yet, pointing at a name that isn't yours.

Hand us a URL and we'll tell you whether it resolves, where it redirects, whether the link and anchor are still on the page, and whether that page is indexed. We can't value a history, can't see traffic, manual actions or reputation — nobody outside Google sees the middle one at all — and we run no link index, so we can't hand you the backlink list at all. That comes from Ahrefs, Semrush or Majestic, or from the marketplace's export: the same index the price came from.

Said once already, up where it mattered. It bears repeating only in this form: the missing index is a limitation before it's modesty.

Questions people ask

Are expired domains safe for SEO?

Buying one violates nothing. Google's expired domain abuse policy describes what gets published on the domain afterwards, so that risk lives in the rebuild, not the purchase. The commercial risk is separate and usually larger: paying for a profile that decayed while the domain was dark.

Can I get penalized for using an expired domain?

Yes, if you repurpose it primarily to rank low-value content on its past reputation. That's the named policy, enforceable by automated systems and by manual action. Using an old name for a new site built for people is explicitly fine, in Google's words.

Do expired domains still work for SEO in 2026?

Sometimes, for a while, and mostly not for the reason a listing implies. What remains is a decayed set of other people's links plus whatever credit Google still passes through them, and a seller can show you neither. Anyone quoting a success rate here invented it.

Does a 301 redirect from an expired domain pass link equity?

Google has never published a figure in either direction, and the widely repeated "90–99%" has no Google source. The answerable questions are whether the old links still exist and whether the old domain's subject relates to where you're pointing it.

How do I avoid buying a domain with a penalty on it?

You can't confirm one from outside. Manual actions appear in the Search Console account of whoever owns the site, and you aren't that person yet. Check the falsifiable things: what the Wayback record shows it published, where the name resolves, how many inbound links still load.

How do you find expired domains with backlinks?

Through drop-catch services, registrar expiry auctions and aftermarket marketplaces, which publish metrics licensed from third-party link indexes. Which is why a listing shows an index figure rather than a live count: the platform isn't checking those links, and neither is the seller.

What you are actually paying for

Strip out the vocabulary the trade has built around this and one sentence is left. You're buying a list of pages, compiled by other people, at a moment you can't verify, about a name that was required to be switched off for weeks before you could reach it.

The list can still be worth money. A defunct competitor in your niche, links from trade press still publishing, redirected to the page that replaces what they sold — an ordinary business decision, and people make it. The word that doesn't belong in it is authority. What you can count instead is how many of those pages still carry the link today: an evening's work, free, doable before you bid, and the only figure in the transaction nobody sold you.