Link rel attributes explained: nofollow, sponsored, ugc and the rest

Most guides to the rel attribute list forty values that nobody uses. For links and SEO, five or six matter. This guide covers those: what each one does, what Google says about it, and where people usually go wrong.

What the rel attribute does

The rel attribute describes how a link relates to the page it sits on. It goes on <a> and <link> tags. Browsers use it to decide how to behave, for example how to open a link safely. Search engines use it as a hint about how to treat the link.

<a href="https://example.com" rel="nofollow">Example</a>

<a href="https://example.com" rel="sponsored nofollow">Example</a>

• href is where the link points.
• rel says how the link relates to your page, so search engines know whether it is paid or not endorsed.
• Anchor text is the words readers see and click.

You can set more than one value. Google accepts a space- or comma-separated list:

<a href="https://example.com" rel="sponsored nofollow">Example</a>

nofollow, sponsored and ugc

These three are the ones link builders care about. Google introduced sponsored and ugc in September 2019, at the same time as it changed nofollow from a strict instruction into a hint.

Google's guidance is simple. Mark paid links sponsored. Mark comments and forum links ugc. Use nofollow when neither fits and you'd rather Google not associate your site with the linked page, or crawl it from your site.

Notice the word Google uses: links marked this way will "generally" not be followed. It's a hint, not a switch. The linked page can still be found through a sitemap or through links on other sites.

Three things follow from that.

Pick the most specific value. If a link is paid, sponsored says so. nofollow says less.

A marked link isn't worthless. It can still send readers to your site and put your name in front of a real audience. Judge a placement on the site and its readers, not only on the attribute.

Paid links are allowed, if they're labeled. Google's spam policies say buying or selling links for ranking purposes is a violation. They also say it's fine to have paid links as long as they carry rel="nofollow" or rel="sponsored". That's why some publishers mark paid content.

What it looks like on a real page

Picture one blog post with three outbound links:

<!-- 1. A paid placement -->
<a href="https://client-site.com/" rel="sponsored">their site</a>

<!-- 2. A link a reader left in the comments -->
<a href="https://someones-blog.com/" rel="ugc">my blog</a>

<!-- 3. A study the author chose because it helps readers -->
<a href="https://example.org/study/">the full study</a>

The third link needs no rel value at all. If you link to something because it genuinely helps your readers, leave it as a normal link.

How people actually use them

The HTTP Archive's Web Almanac looks at millions of real pages every two years. In its 2024 SEO chapter, nofollow appeared on 32.7% of pages, up from 29.5% in 2022. The newer, more specific values are barely used: just 0.4% for sponsored and 0.3% for ugc.

That gap says a lot. Most sites still use nofollow for everything, so don't read too much into which value a publisher chose. What you can control is your own site: use the specific value where it fits.

When we place links for clients, whether a publisher labels a placement depends on the publisher and the country. If a placement will carry a label, we tell you before you approve the site. You can read how that works in our how it works section.

noopener and noreferrer

These two are about behavior, not rankings.

• noopener stops a page opened from your link from reaching back into your page through window.opener.
• noreferrer does the same and also stops the browser sending a referrer, so the site you link to can't see where the visit came from.

MDN notes that target="_blank" now implies noopener behavior on its own, so adding it by hand is harmless but rarely needed.

noreferrer has a side effect worth knowing. If a publisher adds it to your link, visits from that link show up as direct traffic in your analytics. If you're building links partly for referral traffic, that's worth checking.

canonical

rel="canonical" goes in a <link> tag in the page head. It isn't valid on <a> tags.

<link rel="canonical" href="https://example.com/preferred-page/">

It's widely used: the 2024 Web Almanac found canonical tags on 65% of mobile pages and 69% of desktop pages. Google calls it a strong signal that a URL should be the canonical one, but it's not a command. Google may pick a different URL if other signals disagree. Google also recommends absolute URLs. Relative ones are supported but can cause problems later.

hreflang

For sites in several languages or regions, you tell Google about the versions with rel="alternate" and hreflang.

<link rel="alternate" hreflang="en-us" href="https://example.com/us/">
<link rel="alternate" hreflang="de-de" href="https://example.com/de/">
<link rel="alternate" hreflang="x-default" href="https://example.com/">

The rule that trips people up is return links. If page A points to page B, page B has to point back to page A. If two pages don't point to each other, the tags are ignored. Each version should also list itself. x-default is the fallback for visitors whose language you don't cover, and it works best on a language selector page.

You can set hreflang in HTML tags, in HTTP headers or in your sitemap. Google treats the three as equivalent, so pick one and stay consistent. If your site works in many markets, our page on international link building covers how we approach it.

prev and next

You'll still see rel="prev" and rel="next" in older guides. Google says it no longer uses them, although other search engines may. What Google asks for instead is simple: give every page in a series its own URL, link between the pages with normal <a href> links, and give each page its own canonical URL.

Values you can ignore for link building

stylesheet, icon, preload and prefetch matter to developers who build pages. They have nothing to do with link building. MDN keeps the full list if you ever need it.

Six common rel attribute mistakes

1. Using nofollow on your own internal links. You're telling Google not to associate or crawl those pages from yours.
2. Marking every external link nofollow "just in case". Use the specific value where it applies and leave the rest alone.
3. Pointing every paginated page's canonical at page one. Give each page its own.
4. Using relative canonical URLs. Use absolute ones.
5. Adding hreflang without return links. The tags get ignored.
6. Relying on prev and next for Google. It doesn't use them.

A five-minute rel attribute check

Crawl your site with a tool such as Screaming Frog and export the rel values. Then look for four things: comment and forum links without ugc, paid or affiliate links without sponsored, canonical tags with relative URLs, and hreflang pairs that don't point back to each other. Fix those and you've covered most of what matters.

Sources

Checked on 30 September 2026.

• Google Search Central: Qualify your outbound links to Google (updated December 2025)
• Google Search Central blog: Evolving "nofollow" – new ways to identify the nature of links (September 2019)
• Google Search Central: Spam policies for Google web search (updated August 2026)
• Google Search Central: Consolidate duplicate URLs
• Google Search Central: Localized versions of your page
• Google Search Central: Pagination and incremental page loading
• HTTP Archive Web Almanac 2024: SEO chapter (nofollow, sponsored, ugc and canonical usage)
• MDN: rel attribute
• MDN: The anchor element, target attribute