Linking to the source a claim comes from takes nothing away from a page's ranking power. Google's documentation on qualifying outbound links assumes exactly the opposite: you don't need to add any rel value to links you expect Google to fetch and parse normally. In other words, an unmarked link isn't the exception; it's the default behavior.
Even so, many content teams avoid linking out, and when they do, they add nofollow by reflex. This reflex has two separate problems. First, the mechanism it relies on isn't described in Google's current documentation. Second, in the places that genuinely require a decision (ad links, user content, affiliate links), most sites apply the same reflex, so the distinction disappears entirely.
Where the fear of linking out comes from
The fear stems from an old mental model that treats links like plumbing: a page has a fixed "pool of power," every outbound link leaks some of it, so the fewer links you give out, the better. As an extension of this model, a practice called "link sculpting" spread for a while; the idea was to close off some links with nofollow and channel value to the links you wanted.
Current documentation has no counterpart to this model. Google's outbound link document defines rel values not as a valve but as a way to "tell Google about your relationship with the page you're linking to." The document contains no leakage mechanism and no description of value protected by nofollow being passed on to other links.
Some care is needed here: a mechanism's absence from the documentation isn't hard proof that it doesn't exist. But when deciding between two options, working from documented default behavior is more defensible than working from an undocumented mechanism. And the cost of avoidance is concrete: claims become unverifiable and readers are left having to take your word for it.
What Google actually wants for outbound links
The document's logic is simple. If the link is an ordinary editorial reference, you do nothing. If there is a commercial relationship, user-generated content or a destination you don't want to be associated with behind the link, you declare it. Three values exist for this purpose, and they can be combined when needed; rel="ugc sponsored" is a valid combination.
| Situation | Value to use | What the documentation says | Mistake seen in practice |
|---|---|---|---|
| Editorial source that verifies a claim | No value | Normal link, no qualification needed | Adding nofollow by reflex |
| Advertising, paid placement, affiliate | sponsored |
Links that are ads or paid placements | Not marking them at all |
| Comment, forum post, user profile | ugc |
Links within user-generated content | Applying it to editorial links too |
| Destination you don't want your site associated with | nofollow |
Used when the other values don't apply | Thinking it blocks indexing |
| A section of your own site you don't want crawled | Not rel, but robots.txt | Disallow rule for on-site control | Putting nofollow on internal links |
In practice, there are cases where a single link needs more than one value. A link a user left in the comment section that is also known to be a paid placement carries both:
<a href="https://example.com/product" rel="ugc sponsored">product page</a>
<a href="https://example.com/campaign" rel="nofollow sponsored">campaign</a>The nofollow on the second line is added for backward compatibility with systems that don't support the newer values. For Google, sponsored is preferred for marking ad links, but nofollow is also accepted markup for this purpose.
nofollow is a hint, not a directive
The most common piece of received wisdom is that Google ignores a link marked nofollow entirely. Google's own explanation doesn't support this. All three values are treated as hints for ranking purposes, meaning they aren't strict instructions the system is obliged to follow. On the crawling and indexing side, nofollow has also worked as a hint since March 1, 2020; the strict behavior before that date has ended.
This has two practical consequences. First, you can't assume that a page you've only marked with nofollow will never enter Google's index; this isn't the way to keep a page out of the index and was never recommended as such. Second, nofollow isn't a saving mechanism: the documentation nowhere describes value left over from a marked link being distributed to other links.
If you don't want part of your own site crawled, the right tool isn't putting nofollow on internal links but a disallow rule in robots.txt. How to build an internal linking architecture is a separate topic with its own rules; the logic of relevant, in-depth internal linking shouldn't be confused with outbound link markup.
Which sources to link to
Choosing the right rel value is the smaller half of the problem. What really matters is where the link goes. A single principle solves most of it: link to where the claim originated, not to an intermediary publication that relayed it.
Linking to the institution that produced a statistic rather than the site that reported it as news, or to the official document that defines a technical behavior rather than an article interpreting it, is what makes the difference here. A link to an intermediary takes the reader one step further away and lengthens the verification chain.
A practical order that works when choosing a destination:
- The official documentation, standard text or regulator publication that defines the topic
- Peer-reviewed research or an institutional report that openly publishes its methodology
- The primary data source that makes the dataset itself accessible
- If none of the above exist, an industry analysis that explains its own measurement and method
One test is enough for each candidate link: when readers click it, can they verify my claim independently of me? If the answer is no, the link isn't evidence; it's decoration. Inaccessible destinations fail the same test; a page behind a paywall, behind a registration wall or no longer in existence removes the reader's ability to verify.
For anchor text, the standard is to describe the destination. Keyword-optimized anchor text in paid content or guest posts is listed directly as an example of link spam in Google's spam policies. The same discipline applies to links pointing to you; editorial justification is also at the core of the definition of a quality backlink.
Is citing sources really proof of expertise?
A common exaggeration comes up at this point: the claim that outbound links add a "trust score" to a page. In Google's own words, E-E-A-T itself isn't a specific ranking factor; its systems use a mix of different signals that help identify content with good E-E-A-T. Of the four components, trust is defined as the most important, with the others positioned as contributing to trust.
Citing sources plays a more modest and more concrete role in this framework. In Google's content self-assessment questions, when asking whether content would make a reader trust it, clear sourcing, evidence of expertise and information about the author are mentioned together in the same question. So citing sources isn't a score passed to the algorithm; it's an editorial quality that determines how verifiable the content is for the reader.
That's why outbound links don't work on their own. The author background and review record mentioned in that same question are the other trust surfaces that sit alongside the link. Detailed expert author bios show who produced the content, and reviewer information shows who verified it. On topics that directly affect decisions, such as health, finance or safety, all of these layers matter more; Google explicitly says it gives more weight to content aligned with strong E-E-A-T on such topics.
Separating affiliate and sponsored links
The line between editorial links and commercial links is the most neglected side of outbound link policy. Google's definition of link spam groups together buying and selling links for money, exchanging goods or services for links, and sending someone a product in exchange for a review that includes a link. The same policy also covers ranking-passing links in paid articles and native advertising formats.
The distinction here is drawn by relationship, not intent. If there's any compensation behind a link (payment, commission, product, exchange, service), the link is marked sponsored. If there's no compensation and the link is there to verify a claim in the article, it stays unmarked. If you publish paid reviews, you also need to add a visible disclosure for readers alongside the markup; before it's a search requirement, it's a matter of honesty toward the reader.
There's one more easily overlooked clause in the policy: requiring a link as part of a service's terms of use or contract, without giving the content owner the option to qualify it, also counts as link spam. If a plugin, theme or tool tells you "you must link to us in the footer," you have the right to mark that link sponsored or nofollow.
Decision rules
Instead of rethinking these distinctions for every article, you can reduce them to six rules. Applied in order, almost every link is resolved within the first three steps.
- If the link goes to an editorial source that verifies a claim in the article and there's no compensation behind it: add no
relvalue. - If there's payment, commission, product, exchange or service behind the link: add
sponsored. - If a user left the link on the site (comment, forum, profile, listing): add
ugc. You can choose to remove this marking for members who have contributed quality content over a long period. - If you have to include the link but don't want your site seen as associated with the destination (a spam example, a controversial source, a bad practice under review): add
nofollow. - If what you want to block is a section of your own site: use a robots.txt disallow rule rather than link markup. If you want to remove it from the index, use a separate mechanism.
- If the link doesn't give readers a new way to verify something: don't link at all. An unnecessary link that's been marked is still an unnecessary link.
None of these rules includes a numerical target, and that's deliberate. How many outbound links an article has is determined by how many verifiable claims it makes. Five sources aren't too many in a guide with five claims; even one source is too many in a promotional text with no claims. The same logic applies to the broader framework of SEO work: the decision is driven not by the number but by the reasoning behind it.
Frequently Asked Questions
Is there an upper limit on the number of outbound links on a page?
Google's outbound link documentation defines no numerical limit. The practical limit is set by the reader: links so frequent that they interrupt reading break the flow of the text. One source per verifiable claim is a reasonable working measure.
Will the page I link to with nofollow never reach Google?
No. nofollow is also treated as a hint on the crawling and indexing side, not an absolute block. If you want to prevent a page from being indexed, you need to use the mechanisms designed for that purpose.
Should I switch all the outbound links in my old articles to nofollow?
That gets it backwards. Editorial sources should stay unmarked. What to look for in a bulk audit of past content isn't missing nofollow, but affiliate and paid placement links that were never marked.
What should I do if I have to link to a competitor's site?
If the link verifies a claim in the article, leave it as a normal editorial link. Being a competitor isn't in itself a reason to qualify the link; the reasons to qualify are a commercial relationship, user-generated origin or not wanting to be associated with the destination.
