Nofollow, sponsored and ugc are the three values Google reads on a link's rel attribute, and they are not equally important. One of them can get you in trouble if you skip it. One mostly saves you from spam. One is a catch-all. Add the page-level nofollow rule and plain links, and you get five options worth ranking by how much a wrong choice costs.

The ranking below uses Google's own documentation as the scorecard: the outbound links guide, the 2019 announcement that created the newer values, and the spam policies.

The scoreboard

Rank Option Used for Cost of getting it wrong
1 rel="sponsored" Ads, paid and affiliate links High: an unmarked paid link is link spam
2 rel="ugc" Comments, forum posts Medium: invites spammers to your threads
3 Meta robots nofollow Every link on one page Medium: can cut a page's links off from discovery
4 rel="nofollow" Links you don't vouch for Low: a hint either way
5 No rel (plain link) Normal editorial links Low, unless the link was paid

1. sponsored: the one with real stakes

Google's spam policies list buying or selling links for ranking as link spam, including payment in goods or services and a free product given in exchange for a post with a link. The same section says such links are fine once they carry sponsored or nofollow. Google prefers sponsored; its 2019 post said both are treated the same for avoiding a link scheme action.

The rule is lopsided. Google said that putting sponsored on a link that wasn't paid at most means the link might not be credited. Leaving it off a paid link is the error that counts. That asymmetry is why this value sits at the top.

2. ugc: the spam shield

ugc goes on links inside user-generated content, such as blog comments and forum posts. Google's spam policies separately describe forum comments stuffed with optimized links as link spam, and user-generated spam as something site owners should prevent. Marking those links tells Google you didn't place them.

There is a reward mechanism too. Google says you can remove ugc from links by members who have posted high-quality material consistently over time. To stay compatible with services that only know nofollow, you can write rel="ugc nofollow"; values can be separated by spaces or commas.

3. Page-level nofollow: the blunt instrument

<meta name="robots" content="nofollow"> applies to every link on the page. Google's robots meta docs define it as "Do not follow the links on this page", and the value none means noindex, nofollow. It can also be sent as an X-Robots-Tag header.

It ranks this high because it is easy to overuse. Putting it on a page that links to the rest of your site removes that page as a path for Google to discover those links. And only nofollow exists at page level; there is no page-wide sponsored or ugc.

4. nofollow: the catch-all

Google's guide says to use nofollow when the other values don't apply and you would rather Google not associate your site with the linked page. Since September 10, 2019 it has been a hint for ranking, and since March 1, 2020 a hint for crawling and indexing too. Existing nofollow links need no changes, Google said.

Low stakes in both directions, which is why it ranks fourth. It also does not hide a page: linked pages can still be found through sitemaps and other sites. For links within your own site, Google points to robots.txt disallow instead.

No rel value at all is what Google expects for regular links it should fetch and read. There is no dofollow keyword. The only way a plain link goes wrong is when money changed hands, which sends you back to rank 1.

Tallying your own score

To see how the sites linking to you mark their links, the Backlink Finder from seogeoaeo.ai lists the 100 strongest linking sites, each with a link rank from 0 to 1,000, and shows the share of each site's linking pages that are followed rather than nofollow, plus a spam score from 0 to 100. A run costs 70 credits. The figures come from a data provider and are estimates, so they rank sites against each other rather than revealing what Google counts.