Yes, backlinks still affect rankings in 2026. Google’s official documentation, correlation studies and day-to-day link building practice all confirm it. What changed is not how much links matter but which links work: a single link decides almost nothing, and the weight has shifted toward the topical relevance of the donor and how natural the whole link profile looks. Below we go through what the data says, what practice shows, when links don’t help and how they work in the era of AI search.
Key points in 30 seconds
- Backlinks remain one of the key ranking factors, but they are no longer a “magic button”.
- Google has never officially dropped links: PageRank, in a reworked form, is still in use.
- Correlation studies consistently show a link between a quality profile and rankings.
- In AI Overviews and LLM answers, links have an indirect effect through the authority and citation rate of the source.
- Real crowd marketing and careful outreach work; “1,000-link packages” do not.
What studies and Google say
In public, Google downplays the role of links. Company representatives have said more than once that links are no longer among the top three signals and that chasing link volume makes no sense. Still, there has been no statement that link signals were removed: PageRank remains part of the algorithm, just in a heavily reworked form.
What the official documentation says
The most direct confirmation isn’t conference talk but documentation. Google’s guide to Search ranking systems explicitly lists “Link analysis systems and PageRank” among the systems that determine the order of results, and states that PageRank is one of the core systems Google has used since it launched. In other words, links are not an “SEO myth” but a documented part of the algorithm.
The second role of links is technical. Google’s explanation of how Search works describes three stages: crawling, indexing and serving results. Google discovers new pages, among other ways, by following links from pages it already knows. So links affect not only rankings but whether the search engine finds your page in the first place.
The third signal is how seriously Google treats manipulation. A dedicated section of its spam policies covers link spam: “creating links to or from a site primarily for the purpose of manipulating rankings.” Examples include buying or selling links for ranking purposes, excessive link exchanges, automated link-building programs and large-scale guest posting with keyword-rich anchors. If links didn’t influence rankings, Google wouldn’t keep a whole section of rules about them. Nobody installs an alarm system on an empty shed.
The documentation leak and correlation studies
The 2024 leak of Google’s internal documentation showed that the system contains many link-related signals, from source type to trust scores for the donor. This is not “proof” of specific formulas, but it indirectly confirms that links are still taken into account.
Large correlation studies find a stable relationship between the number of quality referring domains and rankings year after year. Backlinko analyzed 11.8 million Google results and found that the #1 result has, on average, 3.8 times more backlinks than results in positions 2–10. Ahrefs’ study shows the other side: 96.55% of pages get no traffic from Google, and there’s a clear correlation between the number of referring domains and a page’s traffic. Of roughly 20 million pages with zero referring domains, only 2,997 get more than 1,000 visits a month, about one in 6,671. For more on what a link profile is made of, see what is link building.
Why correlation isn’t causation
This is worth saying plainly. Correlation only means two values rise together. Pages with many referring domains also tend to be older, better written, owned by well-known brands and receive more clicks. Some of those links may exist precisely because the page was already ranking and got cited more often. So these studies don’t imply “buy X links, get position Y.”
But when official documentation names link analysis as a ranking system and large samples across millions of pages show the same pattern, you can’t ignore it. The sensible conclusion: in competitive niches, links are a necessary but not sufficient condition.
What practice shows
Link building practice gives a more direct answer than any statement. In competitive commercial niches, a young site without external links barely moves above page two or three, no matter how good the content is. And page two of Google is famously the best place to hide something you don’t want anyone to find. As soon as a quality link profile appears, pages start to climb, all else being equal.
The reverse is also true: for low-competition and narrow informational queries, you can reach the top without buying any links, through content and behavioral factors. Where the line runs is covered in detail in can you rank a site without backlinks.
A simple way to check this for your niche: look at the top 10 for your target query and the median number of referring domains. If everyone in the top has dozens of referring domains and you have a handful, no amount of content will close that gap. A step-by-step calculation with an example is in how many backlinks you need to rank.
When links don’t help
When people are disappointed with links, it’s rarely because links “don’t work.” More often they were used where a different factor decides the outcome. Here are five typical situations.
1. Weak or thin content
Links amplify a page; they don’t replace it. Google’s guide to creating helpful content says its ranking systems prioritize content made for people and suggests evaluating pages through the “who, how and why” questions. If your page answers the query worse than competitors do, links will at most push it up to the point where it loses on quality. You can get a weak candidate into the final round, but they still have to do the interview themselves.
2. Search intent mismatch
If the top results for a query are informational articles and you’re promoting a product page, no number of links will convince Google that users want to buy rather than read. First check which page types rank, and only then invest in links.
3. Links Google has neutralized
In its December 2022 link spam update, Google said it uses SpamBrain, its AI-based system, to “neutralize” the impact of unnatural links. The value those links passed is simply lost. So a purchased package of junk links often doesn’t trigger a penalty. It just quietly does nothing. The money is spent, and the profile hasn’t changed.
Don’t confuse this with “toxic” links. As Ahrefs explains, spammy, manipulative and toxic links are different things: spammy links you didn’t build are usually not a problem, and Google’s warnings are about manipulative ones. So don’t disavow everything in sight. Google’s own help page calls the disavow tool an advanced feature and notes that most sites don’t need it.
4. Links that don’t actually exist
A link removed by a moderator, placed on a page blocked from indexing, or not yet crawled by Google has no effect on rankings. A share of “failed” campaigns are simply unverified placements.
5. Technical problems on your site
If the target page isn’t indexed, has a canonical pointing to another URL or returns an error, links lead nowhere. Check the technical state before a campaign; common causes are covered in bought backlinks, no results.
From our practice
The point about “links that don’t actually exist” isn’t theory. In our October 2026 check of 16,500 forums, on 68% of the forums where we could inspect threads, the checked threads contained no external links at all: moderators remove them or don’t let them through. And of 7,872 forums with a visible last-post date, only 58% had new posts in the past three months. That’s why we report the URL of every placement and replace removed links for free: without that kind of control, some “purchased” links exist only in a report.
How long it takes for links to work
Several stages separate placing a link from a change in rankings, and each takes time:
- Crawling the donor page. Google has to find and recrawl the page where the link appeared. According to Google’s recrawl documentation, this can take anywhere from a few days to a few weeks.
- Indexing and counting the link. The donor page has to be indexed, and the link has to be registered in the link graph.
- Recalculating signals for your page. The effect accumulates: a single link is rarely noticeable, while a series of placements over several months is.
- SERP response. Rankings don’t change linearly: long-tail queries usually move first, the main ones later.
The practical takeaway: judging a campaign a week after launch is pointless. It’s like stepping on the scale the morning after your first gym session. A reasonable horizon for a first assessment is a few months of steady work, and six months or more for competitive queries. You can speed up the first stage; see how to speed up link indexing.
The role of links in AI Overviews and LLM answers
AI Overviews and LLM answers have changed not the ranking mechanics but how users see the SERP: the AI answer takes some of the clicks. But the choice of sources for that answer still relies on page authority, and authority is largely built by links.
The chain is simple: quality backlinks raise domain authority → the domain lands in the regular top results more often → models most often pick sources to cite from the top results. There’s a number behind this too: according to an Ahrefs study (March 2026), 38% of AI Overview citations come from pages in the top 10. So rankings in regular search remain one of the most reliable ways to get cited in an AI answer, though not the only one.
The flip side is fewer clicks even at #1. According to Ahrefs’ analysis of 300,000 keywords (April 2025), the presence of an AI Overview correlates with a 34.5% lower average CTR for the page ranking #1. So in 2026 the value of a link isn’t just ranking weight, but also a chance to become a source the AI cites. What Google advises site owners about these features is described in its documentation on AI features in Search. How exactly AI Overviews changes traffic is covered in a separate breakdown, AI Overviews: how traffic is changing, and what to change in your approach right now is in the overview link building trends for 2026.
Does crowd marketing work in 2026
Yes, but only the “real” kind. A crowd link works when it is a relevant comment or answer on a niche platform with a real audience and real discussion. Such a link brings not only a signal for the algorithm but also click-throughs and brand mentions. What separates working crowd links from spam, and how to build them safely, is covered in what is crowd marketing and is it safe to buy backlinks.
Attributes are a separate question. Google recommends marking links in comments and forum posts as rel=”ugc” and paid placements as rel=”sponsored”. At the same time, since 2019 Google has treated nofollow, ugc and sponsored as hints rather than strict directives. So links with these attributes aren’t necessarily “worth zero,” and a natural profile without them would look odd.
The opposite does not work: mass template posting, links from dead forums, “1,000-link” packages with no vetting of platforms. A comment reading “Great post, very helpful! cheap sofas london” in a thread about carburetor repair has never ranked anyone. At best such a profile is useless, at worst it gets you penalized. If you need real crowd links on vetted platforms, we’ll estimate the cost for your project.
From our practice
In our October 2026 check of 16,500 forums, among the 2,765 forums where external links exist at all, 49% are dofollow, 32% nofollow/ugc, 10% mixed, and another 8% go through a redirect. So we tell clients honestly: roughly half of placements, depending on the niche, will be dofollow, and the rest will work as natural profile diversity and a source of click-throughs. A promise of “100% dofollow forum links” is a red flag.
In short, here is what has changed in the role of links in recent years:
| Factor | Before (until ~2020) | Now (2026) | What to do |
| Quantity | The more, the better | Quality matters, not the count | Fewer referring domains, but relevant ones |
| Anchors | Lots of exact-match is fine | Exact-match anchors: penalty risk | Natural anchor diversity |
| Donor relevance | Secondary | One of the main factors | Topical platforms |
| Link velocity | Didn’t matter | A sharp spike looks suspicious | A smooth, steady growth curve |
| Link type | Any link = a plus | Click-throughs and context matter | Real crowd links, outreach, niche edits |
How to test the impact of links on your own site
Instead of taking studies or contractors at their word, check the effect on your own data. Here’s a simple process:
- Record a baseline: rankings of target pages, clicks and impressions in Search Console, number of referring domains.
- Keep a placement log: date, donor URL, target page, anchor. Without it, you can’t match changes to causes.
- Once a month, check whether placements are still live and whether they show up in the Links report in Search Console.
- Compare the trends of pages that received links with pages that didn’t: that’s the closest thing to a “control group” an ordinary site has.
- Account for external events: algorithm updates, seasonality, changes on your site. Otherwise it’s easy to credit links with what an update did.
If, after a few months of steady work, pages with links grow while pages without them stay flat, you’ve got an answer for your niche rather than for an “average” study sample. Our link building plan and anchor text generator tools help you build the schedule and anchor plan for such a test.
The future of links: a forecast without fortune-telling
Links are not going anywhere in the next few years, and there is no sign that Google will drop them completely. The SEO world announces the “death of links” roughly once a year, and every year they somehow survive. But the direction is clear: weight is shifting from “how many links” to “how authoritative and topical the source is”, and E-E-A-T, brand mentions and behavioral data are increasingly mixed in with link signals.
The practical takeaway for 2026: build fewer links but better ones; choose donors by topic, not by a single metric; grow the profile gradually. This is not guesswork but a direct consequence of what the data and practice already show.
Links are still needed in 2026, but the winners are not those with the most links but those with a natural, topical profile. If you want to grow your link profile carefully and without penalty risk, let’s discuss your project and pick a strategy or get a quote for real crowd links right away.
Sources
- Google Search Central: A guide to Google Search ranking systems: Link analysis systems and PageRank
- Google Search Central: In-depth guide to how Google Search works
- Google Search Central: Spam policies for Google web search
- Google Search Central Blog: December 2022 link spam update: SpamBrain
- Backlinko: We Analyzed 11.8 Million Google Search Results
- Ahrefs: Search Traffic Study: referring domains vs. traffic correlation
- Ahrefs: AI Overview citations and the top 10: 38% of citations from the top 10
- Ahrefs: AI Overviews Reduce Clicks
