Yes, backlinks still matter for Google rankings. Google's own ranking-systems guide, last updated in December 2025, says PageRank is still part of its core ranking systems. But that single fact does not tell you whether link building still works for your budget this quarter, and it does not mean every new link will move a page up. The honest answer sits between the two camps you usually hear: not "links are dead," and not "links are still a top three factor like it's 2016."
Here's the grade on the overall question: strongly supported. Google continues to use link-based systems, and a large recent study found a positive, if weak, relationship between backlinks and rankings. What isn't supported is the idea that more backlinks reliably produce higher rankings, or that there's a magic number of links you need to rank at all. If you're trying to settle backlinks ranking factor 2026 debates inside your own team, or just deciding whether link building still work for your budget, the real question isn't whether backlinks matter. It's when earning more of them is a defensible place to put your money.
Why so many people think backlinks stopped mattering
This debate didn't come out of nowhere. Back in 2016, Google's Andrey Lipattsev named links as one of the top three ranking factors, alongside content and RankBrain. That quote got repeated for years, and a lot of SEO advice still treats it as current.
It isn't. By September 2023, Google's Gary Illyes told a Pubcon Pro audience in Austin that people overestimate the importance of links, according to reporting from Search Engine Land. He said rankings without backlinks are possible. A year earlier, in November 2022, John Mueller had already signaled the direction of travel on a Search Off the Record podcast episode, saying links probably wouldn't stay as big a factor as they sometimes were. Then in April 2024, Illyes went further at SERP Conf, saying Google needs very few links to rank pages and that "over the years we've made links less important," per Search Engine Roundtable's coverage.
Put those comments together and you get a real, documented shift in how Google's own staff describe the role of links. What you don't get is a number. None of these statements say how much less important links have become, and none of them say links stopped working. A signal can still be in active use while carrying less relative weight than it did a decade ago, and that's the most accurate way to read what Google has actually said.
What the ranking studies actually show
Industry data backs up the "still relevant, not dominant" read, but you have to read the numbers correctly.
The most current large study is from Ahrefs, published in January 2025. Patrick Stox analyzed 1,000,000 of the highest-volume US keywords and found a correlation of 0.248 between the number of backlinks a page had and its ranking, and 0.255 for the number of distinct linking websites. Stox describes both as weak correlations, and he's right to. A correlation around 0.25 means there's a real, positive relationship, not that backlinks explain a quarter of what determines rankings. Informational queries showed stronger backlink correlations than commercial or transactional ones, and higher search-volume terms showed stronger correlations too, which matters if your pages mostly target a narrower, lower-volume niche.
Go back further and the headline numbers get bigger, which is exactly why they get repeated past their shelf life. The 2020 Backlinko and Ahrefs study of 11.8 million search results found that the number-one result had, on average, 3.8 times as many backlinks as pages in positions two through ten. That statistic gets quoted constantly, almost always without its most important footnote: about 95% of the pages in that dataset had zero backlinks and were excluded from the backlink-count analysis entirely. The 3.8x figure describes a filtered subset of already-ranking pages from 2020, not a target you can build toward in 2026.
Then there's the 2024 research from Internet Marketing Ninjas, covered by Search Engine Land, looking at commercial-intent searches specifically. An initial sample of 200 keyword phrases and 1,113 websites found more than 96% of the sites had over 1,000 backlinks from distinct domains. A follow-up expanded that to 2,000 phrases and 9,337 domains, and found 94% of sites had at least 500 distinct linking domains. That's a real pattern among businesses competing for commercial searches. It is not proof that 500 links is a requirement, because the study only looked at sites that were already ranking. It can't tell you how many sites tried to rank with fewer links and failed, or how many ranked with fewer links and succeeded, because that denominator isn't in the data.
The one study here that goes beyond correlation is Ahrefs' 2021 disavow experiment. Stox disavowed 3,476 links across three of the company's own blog posts, leaving the file active for four weeks before removing it. Estimated traffic to the SEO pricing page fell 13.3% during the disavowal and recovered to 99% of its prior level once it was removed. A YouTube searches page fell about 18% and only partially recovered. A Bing searches page actually rose during disavowal, driven mostly by one search term. That's a genuine intervention, not just an observed pattern, which makes it more informative about cause and effect than any correlation study. It's also just three pages on one site over a few weeks, with no control group, so treat it as supporting evidence rather than a settled answer. Stox himself has warned readers not to go disavow their own links based on this.
What the evidence doesn't prove
It's worth being precise about what these numbers can't tell you, because this is where a lot of budget decisions go wrong.
A correlation between backlinks and rankings doesn't separate cause from effect. Pages that already rank well tend to get discovered and linked to more often, simply because more people see them. Search Engine Land's coverage of the 2024 commercial-query study makes this point directly: ranking well can itself attract links, which means some of what looks like "links caused the ranking" is really "the ranking (or the popularity behind it) caused the links."
A page's own backlinks and its website's backlink profile are also two different measurements, and mixing them up explains a lot of the apparent disagreement you'll see across different statistics. A single page can rank with very few direct links while sitting on a site that has thousands of linking domains overall. The "do I personally need links" question and the "does my site need an established link profile" question have different answers.
None of the studies here establish a minimum link count, a predictable return on a specific link-building campaign, or a time-to-results window. If someone tells you a page needs exactly 500 backlinks to rank, or that buying three times more links than your competitor guarantees you'll pass them, that claim goes well beyond anything in the public research. The honest position is that backlinks are one input in a multi-signal system, and the system doesn't publish its weights.
What Google confirms today, and where it draws a hard line
Google's current documentation settles the "is this still a thing" question more clearly than the importance question. The ranking-systems guide, last updated December 10, 2025, explicitly states that PageRank "continues to be part of our core ranking systems." That's about as direct a confirmation as you'll get from Google on anything algorithmic.
What Google is less generous about is tolerating manipulation. Its December 2022 link spam announcement describes SpamBrain's ability to detect sites that buy links and sites set up to pass outgoing link credit for payment. When Google identifies that kind of unnatural link activity, it neutralizes the credit the link would otherwise have passed, and its spam policy documentation says sites that violate these policies can rank lower or be removed from results entirely. There's an important nuance buried in that guidance too: once a link-spam update removes a benefit a site was getting from manipulative links, simply taking the links down afterward doesn't automatically restore what was lost. The credit doesn't come back just because the evidence of manipulation does.
That distinction matters for how you think about spend. A link-building campaign can cost real money and deliver no durable ranking benefit, not because links don't work, but because the specific links purchased never carried legitimate credit to begin with, or had that credit stripped out. Paying for placements is not the same thing as earning a signal Google will actually count.
The verdict for your link-building budget
Here's the practical read for a marketing lead weighing this. Don't cut link-building spend to zero just because someone told you links aren't top three anymore. That quote is outdated and was explicitly walked back by Google's own staff, and the underlying mechanism Google describes is still active. But don't keep funding link acquisition on autopilot just because a competitor has more backlinks than you do, either. A backlink gap next to a ranking gap is a hypothesis worth testing, not a diagnosis. The competitor might also have stronger content, more site history, or simply more existing popularity, and any of those could explain the gap on its own.
A defensible middle position looks like this: keep earning genuine editorial links where the underlying content is good enough to deserve them, and treat that as one input into an organic strategy rather than the whole strategy. Don't use a backlink gap as an excuse to skip the harder question of whether the page itself actually serves the query. And when you review a link-building program, look past delivery counts (how many links got placed this quarter) toward outcomes: did the pages in question move in rankings or clicks that you can plausibly attribute to something other than noise. A vendor who reports only "links delivered" isn't reporting on results.
What would change this verdict
This read could shift with better evidence. If Google ever documented that external links no longer participate in ordinary ranking at all, that would be a real change, not a continuation of the gradual-deprioritization story it has been telling since 2022. On the other side, a larger set of independently replicated interventions, closer in design to Ahrefs' disavow test but run across more sites and more query categories, would sharpen how confident anyone can be about the actual size of the effect. Right now, the honest summary is that backlinks are part of Google's ranking machinery, and the leap from that fact to a guaranteed result from your next batch of acquired links is exactly the part the evidence doesn't support.
