Do social media links and social signals help SEO?
Links from social posts are real links, but almost all carry nofollow or ugc and pass little or no ranking credit. Google staff have said for years that likes and shares are not ranking signals. The value is indirect, and social pages are now sources that AI assistants cite.
In short
- Google's Gary Illyes said in 2017 that most social media links count, for PageRank, "as much as a single drop in an ocean".
- John Mueller said in 2021 that Google does not use likes as a ranking factor, and Matt Cutts said in 2014 that follower and like counts were not in the algorithm.
- Social distribution helps indirectly, because people who see a piece of content may later link to it, write about it or search for the brand.
- LinkedIn was in the top five cited domains on AI Mode, AI Overviews, Perplexity and ChatGPT in Peec AI's March 2026 analysis of 30 million citations.
- A US Federal Trade Commission rule announced in August 2024 bans buying or selling fake followers and views, so bought social signals are a legal risk as well as useless.
Social media links count as backlinks only in the loosest sense. They are links from another site, but nearly all of them carry a nofollow or ugc attribute, and Google has said repeatedly that likes, shares and follower counts are not ranking signals. If you post your URL on ten platforms, your rankings do not move. Anyone selling “social signals” as a ranking service is selling something Google says it does not count.
Social still matters to off-page work for two reasons. It is how content reaches the people who do link, and public social pages are now among the sources AI assistants cite most. Those are quite different kinds of value from the one the phrase “social signals” was coined to describe, and much of the confusion on this subject comes from treating them as one. This lesson separates the three things that get bundled under that label and says what each is good for.
Three things people mean by social signals
When someone says social signals help SEO, they could mean the link on a profile, the links inside posts, or the engagement numbers attached to those posts. The three behave differently, so it helps to set them side by side before looking at what Google has said about any of them.
| What | Example | Direct ranking value | What it is good for |
|---|---|---|---|
| Profile links | The website field on a LinkedIn or X profile | None to speak of. One nofollow link per platform | Helping search engines and people connect the profile to your brand |
| Links in posts | A post that links to your new study | None to speak of. Nofollow or ugc | Clicks, and reaching people who may link later |
| Engagement counts | Likes, shares, followers, views | None, according to Google | A rough sign that content is getting attention |
The attribute matters because of how Google treats it. Google’s documentation on qualifying outbound links says links marked nofollow, sponsored or ugc “will generally not be followed”. Link attributes explains each one. The consequence for social links is that the first two rows of the table describe links that exist and can be clicked but that Google generally does not follow, which is why the right-hand column lists benefits that have nothing to do with passing ranking value. A profile link helps tie the profile to your brand, a post link sends readers, and neither is counted the way an editorial link from another site would be.
What Google has said
Search Engine Journal collected the statements in a 2022 review of the question, and its verdict was that social signals “seem to be a thing of Google’s past”. The statements it records run across several years and several Google spokespeople, and they all point the same way. Matt Cutts said in 2014 that signals such as the number of followers or likes were not included in the algorithm. John Mueller said in 2015 that social signals do not directly help in organic rankings. Gary Illyes said in 2017: “PageRank wise most social media links count as much as a single drop in an ocean.” And John Mueller again, in 2021: “Sorry, we don’t use likes as a ranking factor.”
The same review notes that Cutts said in 2010 that Google had used social links as a signal at that time. So the position changed more than a decade ago and has been consistent since. That history is worth knowing because it shows where the idea came from: there was a period when a Google spokesman said social links were used. A claim that Google counts social signals has only the 2010 remark to lean on, and has to set aside everything said from 2014 onwards.
There are practical reasons for Google’s position. Social platforms restrict what crawlers can see, counts can be bought, and a like is a weak statement compared with an editor choosing to link. We found no primary Google document that sets out this reasoning, so treat it as the usual explanation and not a quotation. It does hang together, though. A signal that Google cannot reliably read, that anyone can purchase, and that costs the person giving it a single tap would be a poor basis for ranking pages, whatever its appeal as a measure of popularity.
Why shares and rankings still move together
Pages with many shares often rank well, and that is easy to read as cause. It is correlation. Content that is useful or newsworthy gets shared, and the same content gets linked, mentioned and searched for, so the shares and the rankings rise together because both follow from the quality of the thing published. The shares did not cause the ranking. The indirect route is real, though, and it is the reason to bother with social at all. It runs in four steps.
- You publish something worth citing, such as original data or a tool.
- You post it where writers in your field spend time.
- Some of them see it, and a few link to it, quote it or include it in a newsletter.
- More people search for your brand by name.
Steps 3 and 4 are off-page signals that search engines do use. Links are counted directly. Brand searches and clicks feed Google’s click-based systems, which our lesson on off-page reputation signals covers. Social is the delivery channel for both, which means its contribution depends on what it delivers. A post with thousands of likes that leads to no link, no mention and no extra brand searches has, on Google’s own account of its systems, done nothing for rankings, while a post seen by a handful of the right writers may lead to several links.
This is why practitioners treat social as the distribution layer for linkable assets and for digital PR. Journalists find stories on social platforms, so a campaign that nobody posts about has fewer chances of being picked up. The asset or the story still has to be good enough to earn the link once it is seen. Social only decides whether the people who could link ever come across it.
What changed with AI assistants
In classic search, a social post was a nofollow link at best. In AI answers, the post itself can be the source, and the citation data from 2025 and 2026 shows social and community platforms near the top of the lists. Peec AI analysed 30 million citations in the US in March 2026. Its overall top ten cited domains were Reddit, YouTube, LinkedIn, Wikipedia, Forbes, G2, Yelp, Facebook, Medium and TechRadar. Four of those are social or community platforms. LinkedIn was in the top five on Google AI Mode, AI Overviews, Perplexity and ChatGPT. Facebook was in the top five on Google’s two surfaces. Peec also notes that Facebook blocks Perplexity’s bots, which is a reminder that a platform can only be cited by an assistant that is allowed to read it.
A second study points the same way for one platform in particular. Semrush’s study of 230,000 prompts from July to October 2025 found LinkedIn cited in about 15% of AI Mode responses, the highest of any domain in that sample. For a business audience that makes LinkedIn the obvious place to put findings you would like an assistant to repeat, with the caveat that the figure describes one Google surface over a few months.
Two cautions apply before building a plan on these numbers. These are vendor studies, and Peec publishes no percentage for each domain, so its top ten tells you the order and nothing about the size of the gaps. Profound’s analysis of 680 million citations, covering August 2024 to June 2025, shows that even the leading domains hold single-digit shares of all citations, which means being a top-cited platform still leaves the great majority of citations going elsewhere. And the mix changes fast: how AI search chooses sources records two sharp falls in Reddit’s share within a year. It is also reported that public posts from Instagram professional accounts became indexable by search engines in July 2025. We did not verify that against Instagram’s own documentation, so take it as reported.
The practical reading is that what you and other people say about your brand in public posts can now be retrieved and quoted. That is a reason to keep company and staff profiles accurate and to publish substance natively on the platform, not only a link, because an assistant can quote a finding that is written out in the post and has nothing to quote from a bare URL. It is not evidence that follower counts or likes influence which brands an assistant names. None of the figures quoted above concerns engagement, so the old idea of social signals gets no support from them.
What is worth doing, and what is not
The useful work starts with claiming your profiles and keeping them consistent: the same name, description and URL everywhere. This is hygiene that helps search engines treat the profiles and the site as one entity. After that, post your assets where writers are. For most business topics that means LinkedIn and X. For consumer topics it may be YouTube, covered in YouTube and AI visibility, or Reddit, covered in Reddit for SEO and AI visibility. Write for the platform when you do, since a LinkedIn post that states the finding in full can be cited by an assistant and a bare link cannot.
Then follow up on what social produces. When a post leads someone to write about you without a link, that is an unlinked mention you can ask to have linked, and it would not have existed without the post. Track the outcomes that count, which are new referring domains, mentions and brand searches after a campaign, not likes. Measuring a campaign by its engagement numbers repeats the original mistake, because it rewards the one thing in the table above that Google says it does not use.
Buying engagement sits at the top of the list of things not worth doing. Bought likes, shares and followers do nothing for rankings because the counts are not used. They are also a legal risk in the US. The Federal Trade Commission’s final rule on fake reviews and testimonials, announced on 14 August 2024, prohibits selling or buying fake indicators of social media influence, such as followers and views, where the buyer knew or should have known they were fake. So the purchase has no SEO upside and carries a regulatory downside, which is as clear a case against a tactic as this subject offers.
Social bookmarking and profile link packages fail for a related reason. Sellers offer hundreds of profile links or bookmarks as “social signals”. These are the same nofollow links at volume, usually created by software. See automated links for how those packages work and why they are ignored. The same thinking applies to link reports. If a vendor’s report pads its total with profile and post links, the number tells you nothing, so count editorial links from other sites separately and judge the vendor on those.
Where to go next
The strongest argument in this lesson is that social work pays off through mentions, links and brand searches, so the natural follow-up is the evidence on mentions as a signal in their own right, which is set out in brand mentions and citations. Terms such as nofollow, ugc and entity are defined in the glossary, which is worth keeping open if the attribute names in the first section were new to you.
Common questions
Do social media links count as backlinks?
Technically yes, since they are links from another site. In practice most are marked nofollow or ugc, and Google's documentation says links with those attributes will generally not be followed. They are worth having for traffic, not for ranking credit.
Are likes and shares a Google ranking factor?
No, according to Google. John Mueller said in 2015 that social signals do not directly help organic rankings, and in 2021 that Google does not use likes as a ranking factor.
Why do pages with many shares often rank well?
Because the same things cause both. Content that people share also tends to attract links, mentions and brand searches. The shares are a sign of that attention and not the cause of the ranking.
Is social link building worth doing?
As distribution, yes. Posting a linkable asset where journalists, bloggers and newsletter writers will see it is how it earns real links. As a way to collect ranking links from profiles and posts, no.
Do AI assistants use social media content?
They cite it. Reddit, YouTube, LinkedIn and Facebook all appeared among the ten most cited domains in Peec AI's March 2026 analysis. That makes public social posts a source for AI answers, which is a different thing from a ranking signal.

