What the attributes are
These are values of the rel attribute on an HTML link, and they tell a search engine something about the relationship between the linking page and the destination. There are three relevant to link building.
rel="nofollow"— the original. It signals that the linking site does not vouch for the destination and does not want the link treated as an endorsement.rel="sponsored"— marks a link created as part of an advertisement, a sponsorship, or any other compensated arrangement, including affiliate links.rel="ugc"— marks links in user-generated content: comments, forum posts, user profiles, anything the site's own editorial staff did not write.
They can be combined, as in rel="nofollow sponsored", which many publishers do for safety. A link with none of these attributes is conventionally called a followed or dofollow link, though dofollow is not a real attribute — it is just the absence of the others.
All of this is markup on somebody else's page. You do not control it, you cannot negotiate it in any way that is safe to negotiate, and in most cases you will not know which you are getting until the coverage appears.
2005: why nofollow was invented
Nofollow was introduced in 2005, jointly announced by Google and other major search engines, and it was built to solve one specific problem: comment spam.
Blogs had become enormous, and every blog comment form was an open invitation to insert a link into somebody else's page. Automated tools filled comment sections with links to pharmaceutical and gambling sites at industrial scale. Publishers were spending real time deleting them, and the underlying incentive was that the links counted.
Nofollow removed the incentive. Blogging platforms began applying it automatically to comment links, and it worked — not by stopping the spam, which continues to this day, but by making it pointless. That original purpose explains why blog comment links are graded obsolete on this site. They were killed twenty years ago and the industry that sells them never quite noticed.
Nofollow then spread far beyond its intended use. Wikipedia applied it to all external links. Major news organizations applied it broadly. Social platforms applied it to everything. Within a few years, a large share of the most authoritative sites on the web had made themselves structurally incapable of passing a followed link, which is the situation we still live in.
2019: hints, and two new attributes
In 2019 Google announced two changes, which took effect for ranking purposes in March 2020.
First, it introduced sponsored and ugc as more specific alternatives to nofollow. The reasoning given was that nofollow had been made to carry too many different meanings — paid link, untrusted user content, and simple non-endorsement are three genuinely different situations, and collapsing them into one attribute threw away information.
Second, and much more significantly, Google said it would treat all three attributes as hints rather than directives. A directive is an instruction that must be obeyed. A hint is information to be weighed alongside everything else. In plain terms: Google reserved the right to consider a nofollowed link anyway.
The stated rationale was that link data is useful and a blanket exclusion discarded good signal. That is a reasonable explanation. It is also worth noticing what the change did for Google, which was to remove a constraint that publishers had been imposing on it unilaterally for fourteen years.
Publishers were told they did not need to change existing markup. Most did not. The practical result is that a large share of the web's nofollow attributes were applied for reasons that predate the distinction the new attributes were meant to capture.
The honest part: nobody outside knows what hint means
This is where most articles on the subject start inventing, so let me be precise about what is known and what is not.
Known: the attributes exist and are documented. Google stated it treats them as hints. That statement is public and has been repeated.
Not known, and not knowable from outside: when a hint is followed, when it is ignored, whether the treatment differs by attribute type, whether it differs by the trustworthiness of the linking site, whether a nofollowed link passes ranking credit at full strength, reduced strength, or none. No experiment run from outside can isolate this variable, because you cannot hold every other input constant across a live search index.
Anyone telling you that nofollow links now pass 40% of the equity of a followed link has made that number up. Anyone telling you nofollow links are worthless is asserting something Google has explicitly contradicted. The honest position is that we know the policy and not the implementation.
What I would say from experience, hedged appropriately: the distinction has mattered less each year. Coverage in a major national outlet that nofollows everything routinely produces movement that a followed link from a mid-tier blog does not. Whether that is the link being counted, the brand mention being counted, the traffic, the subsequent citations by other outlets, or all four is not separable. It does mean that refusing coverage because it will be nofollowed is a decision I have never seen pay off.
How much nofollow is normal
A number, from real data rather than assertion. On the twenty-year profile used throughout this site — 1,301,839 links, 22,260 referring domains, none ever bought — 16.4% of links carried nofollow.
That figure is worth holding onto for two reasons. First, if you are auditing a profile and find a similar proportion, that is what an entirely organic profile looks like; it is not a problem to fix. Second, it is a floor rather than a ceiling for what to expect from certain kinds of work. A digital PR campaign landing in national press, or a program built on forum participation and user-generated platforms, will show a far higher share, and that is a description of where the links came from rather than a quality judgment.
Two profiles with identical follow ratios can be completely different in quality, and two profiles with wildly different ratios can be equally good. The ratio tells you about link sources. It does not tell you about link value.
What to do about all of this
Practical guidance, in rough order of how often it comes up.
Do not filter prospects by follow status. Excluding sites that nofollow removes Wikipedia, most major news organizations, most large forums and most social platforms from your target list. Those are frequently the best placements available to you, and the associated traffic, brand exposure and downstream citations are real regardless of the attribute.
Never ask a publisher to remove nofollow. It marks you as someone chasing a link rather than coverage, it is a request the person you are asking usually cannot grant, and on a paid arrangement it is asking them to conceal a paid relationship. I have watched this single request end otherwise successful media relationships.
Mark your own outbound links correctly. This is the part fully within your control and it is frequently neglected. Use sponsored for anything compensated, including affiliate links. Use ugc on comments and user submissions. Selling followed links from your own site is on the wrong side of the published policies in both directions.
Record the attribute, do not report on it. Track it so you can answer the question when asked. Do not build a report around a followed-link count, because it implies a precision about how the links are treated that nobody outside Google possesses.
Lesson 9.5 of the course, on why dofollow matters less than it did, develops this argument in full.
Questions
Do nofollow links help SEO?
Google says it treats nofollow as a hint rather than a directive, meaning it may consider such links. What it does in any specific case is not observable from outside, and anyone quoting a percentage of equity passed has invented it. What is clearly real is the traffic, the brand mention, and the downstream coverage a good nofollowed placement generates. Those alone justify pursuing it.
What is the difference between sponsored and nofollow?
Sponsored specifically marks a link created through payment, advertising or an affiliate arrangement. Nofollow is the older, general-purpose attribute meaning the site does not vouch for the destination, for any reason. Since 2020 sponsored is the correct attribute for compensated links; nofollow still works for that purpose, but it carries less information about why it was applied.
How much of my link profile should be nofollow?
There is no target. On a twenty-year profile where no link was ever bought, 16.4% of links were nofollow, which gives you a reference point for an organic profile. A heavy digital PR or community-driven program will run much higher and that is fine. The ratio describes where your links came from, not how good they are.
Should I use nofollow on my own outbound links?
Not on ordinary editorial links to sources you are citing — those should be plain followed links, and linking out to good sources is normal publishing behavior. Use sponsored on anything compensated, including affiliate links, and ugc on comments and user submissions. Blanket-nofollowing every outbound link is a habit from an older era that gains you nothing.