Do Backlinks From Sites on the Same IP Address Count?

Yes, they count. A shared IP address is not a penalty signal on its own — most of the web sits on shared hosting or behind a CDN, so millions of unrelated sites share addresses. The IP only becomes a problem when it appears alongside other evidence that the linking sites are all controlled by one person.

Where the "class C IP" idea came from

Ask an SEO who was working in 2008 and they'll tell you links need to come from "unique class C IPs" — meaning the third block of an IPv4 address (the 198.51.**100**.7 part) should differ between your linking sites.

It wasn't nonsense at the time. In the early 2000s, running a website usually meant renting a server or a block of addresses, so a run of sites sharing a subnet genuinely suggested one owner. Link-network vendors started advertising "unique C-class" inventory, the phrase stuck, and it has outlived the conditions that made it meaningful.

Why the signal stopped working

Three things broke it:

Shared hosting became the default. A budget hosting plan can put hundreds or thousands of unrelated sites on one IP. Your dentist, a fan wiki and a Shopify store can all answer on the same address with no relationship whatsoever.
CDNs put a proxy in front of everything. If a site uses Cloudflare, Fastly or similar, the IP you resolve belongs to the CDN's edge network, not the site. Enormous numbers of completely unrelated domains resolve to the same handful of addresses. The IP tells you which CDN they chose, and nothing else.
Cloud platforms pool addresses. Sites on Vercel, Netlify, AWS or Squarespace share infrastructure by design.

A signal that fires on most of the legitimate web isn't a useful way to detect manipulation. Google's search advocates have said as much repeatedly over the years — shared hosting is normal, and using it doesn't disadvantage a site.

What actually gets networks caught

The mistake is thinking of the IP as the thing being detected. It isn't. What gets a private blog network caught is the footprint — the accumulation of coincidences that no set of genuinely independent sites would produce.

IP is one row in a long table:

Signal	Harmless alone?	Suspicious in combination
Shared IP or subnet	Yes — shared hosting is everywhere	Yes, if paired with the rows below
Same registrar	Yes — a handful dominate the market	Yes, alongside identical registration dates
Identical WHOIS contact	Sometimes	Strongly
Same CMS and theme	Yes — WordPress runs a large share of the web	Yes, if the theme is unmodified across all of them
Same Analytics or AdSense ID	No	Very — this is a direct ownership link
Sites linking to each other in a ring	No	Very
Same writing style, publish cadence, image source	Sometimes	Yes
All outbound links point to the same small set of targets	No	Very

Notice that the strongest rows have nothing to do with hosting. A shared tracking ID or a closed link ring identifies common ownership far more reliably than an IP does, and neither is hidden by buying scattered hosting. This is the flaw in every "diverse IP network" product ever sold: the IP was the easiest signal to spoof and the least important one.

So should you check a linking site's IP?

Occasionally, and never as your first test. It's a cheap sanity check when something else already looks off — a site that offered you a link out of nowhere, or a swap partner whose content reads like it was produced in bulk.

You can resolve one from any terminal:

nslookup example.com

Reverse-IP lookup tools will then show you what else answers on that address. Read the output carefully: if the result is a CDN range, the list is meaningless. And if a site you were offered a link from shares an IP with fifty other sites that are all in unrelated niches but all published the same week, you've learned something — but what you've learned is about the content, not the address.

The far better checks are the ones a reverse-IP lookup can't give you: does the site have real traffic, does anyone else link to it, does a human being clearly write it, and would you be happy for a client to see your link there. Our guide to vetting a link swap partner walks through that in order.

What this means if you exchange links

Two practical points.

Don't reject a partner for sharing a host. If you rule out every site on shared hosting you'll rule out a large share of the independent web — exactly the small, real, topically-relevant sites that make good partners.

Do watch for one person offering you several "different" sites. This is the version of the problem that actually reaches you. If the same contact offers links from four domains, those four links are worth roughly what one link is worth, for the same reason multiple links from the same site have diminishing returns — it's one endorsement, not four. And if that person's network is later identified as a link farm, you're holding four links from it rather than one.

This is the structural argument for trading with strangers rather than assembling a stable of friendly sites. Backlinkster matches you with a different site owner for each 1-for-1 in-content swap and verifies the placement by code, so the links you accumulate come from genuinely separate parties — which is the property that matters, rather than the address they happen to resolve to. See the plans.

Frequently asked questions

Does Google penalise sites on the same IP address? No. Shared hosting and CDNs mean a large portion of the web shares addresses with unrelated sites, so the IP by itself carries almost no information about ownership. Penalties follow evidence of a manipulative network, which is built from many signals together.

Do backlinks from the same IP pass link equity? Yes, the same as any other link. The value is determined by the quality and relevance of the linking page, not by its hosting. Several links from genuinely separate sites that happen to share a host still count as separate referring domains.

Is "unique class C IP" still a useful requirement for link building? Not really. It was a reasonable proxy for independent ownership in the early 2000s and stopped being one once shared hosting and CDNs became standard. Any vendor still selling "unique C-class links" is selling the easiest signal to fake.

How do I check what IP a website is on? Run nslookup example.com or dig example.com from a terminal, then use a reverse-IP lookup to see what else resolves there. If the address belongs to Cloudflare or another CDN, the neighbour list tells you nothing useful.

If two of my own websites share a host, can I link between them? Hosting isn't what makes that risky. A small number of genuinely useful links between related sites you own is normal; systematic cross-linking designed to pass authority is the problem. See can I link between my own websites.

Should I pay extra for hosting on a dedicated IP for SEO? No. There's no ranking advantage to a dedicated IP. There can be non-SEO reasons to want one, but search performance isn't among them.

The bottom line

Same-IP backlinks count, and treating a shared address as a red flag will mostly cause you to reject legitimate small sites. The question worth asking about any group of linking sites isn't "do they share a subnet?" but "are these genuinely different people?" — because that's the question search engines are trying to answer too, and they have far better evidence than an IP to answer it with.