# Backlinks still matter for SEO, but many count for nothing - Episode: P2E01 (Part 2: Link building) - YouTube: https://www.youtube.com/watch?v=-v2qsHZToYI - Page: https://irongoo.com/videos/backlinks - Duration: 11:29 - Published: 2026-11-05T17:15:00+05:00 Learn why backlinks still influence Google rankings and why many paid links deliver no ranking value. Find out when your site needs links and how brand mentions relate to visibility in AI answers. ## What this episode covers - What Google's court record says about anchors, relevance and PageRank. - How SpamBrain can neutralise unnatural links without a penalty or warning. - Why a strong link profile still leaves a weak page needing work. - What research shows about brand mentions, earned media and AI visibility, with correlation kept separate from causation. - How rankings may connect backlinks to AI visibility indirectly. - How to check the relevant links behind competing pages before setting a budget. ## Key moments - [00:00 Intro](https://irongoo.com/videos/backlinks?t=0) - [00:45 From the factor to one input](https://irongoo.com/videos/backlinks?t=45) - [01:47 What the court record shows](https://irongoo.com/videos/backlinks?t=107) - [03:45 The bar has risen](https://irongoo.com/videos/backlinks?t=225) - [05:46 Where AI answers look instead](https://irongoo.com/videos/backlinks?t=346) - [08:44 Does your niche need links?](https://irongoo.com/videos/backlinks?t=524) - [11:00 Wrap up](https://irongoo.com/videos/backlinks?t=660) ## Transcript ### Intro Here's a game you can play tonight. Find the loudest person online saying backlinks are dead and you should just write great content. Now put their own site into any backlink tool. Thousands of linking domains, usually. Some earned, some from partnerships, plenty from guest posts and the odd paid placement. Nothing wrong with any of that. It's just a strange sermon to preach from on top of a very large pile of links. So let's skip the opinions and read the record. Links still matter. They matter differently than they did ten years ago, and most of the money spent on them is wasted. Both are true, and by the end you'll know where links sit today and where AI answers look instead. ### From the factor to one input Start with Google's own story, because it keeps changing. In 2016 a Google search strategist named the top three ranking signals in a live Q and A. Links were on the list, sharing first place with content. By 2023 a Google analyst was telling a conference that links hadn't been a top three factor for some time. In April 2024 he went further: we need very few links to rank pages. Within days he posted that he shouldn't have said it. Twice, for emphasis. Follow the arc. Google went from links are a top signal to links matter less than you think. It never got to links don't count. And the one time the message came close, its own analyst took it back. Why would Google play links down in public? Our read, not Google's: every time Google says links matter, the link selling industry gets a fresh sales slide. Playing them down costs nothing. Policing them costs a lot. So read public statements as messaging, and look for harder evidence. ### What the court record shows Press quotes are one thing. Court exhibits are another. In the US antitrust case against Google, one trial exhibit holds notes from a February 2025 call with a Google engineer about how ranking works. It names three fundamental signals and calls them the A B C. A is anchors: a page pointing to another page, so links and the words in them. B is body, the words on the page itself. C is clicks, historically how long someone stayed before bouncing back to the results. Together they make topicality, Google's call on how relevant a page is to a query. So links feed relevance. They feed quality too. The same exhibit describes PageRank as a signal about distance from a known good source, used as an input to the quality score. The judge's September 2025 remedies opinion calls it a key quality signal. Not bad for something people have been burying since 2016. Now the part link sellers skip. In the same hearing, a question put to Google's expert framed it like this: most of Google's quality signal comes from the webpage itself. So a strong link profile on a weak page gets you a well connected weak page. Links turn up in one more place. The remedies opinion describes a popularity signal that, according to two exhibits, uses Chrome visit data and the number of anchors to promote well linked documents. So being linked widely may also make a page look popular. Two verdicts come easy, then. Links are dead: placebo, because the record names them in relevance and in quality. Links are everything: also placebo. They're one input, and most of the quality call comes from the page. What the record doesn't give us is weights. Nobody outside Google knows how anchors stack up against body text, or PageRank against everything else. So when someone tells you links are exactly forty percent of ranking, ask which exhibit they read that in. We work with what's named and test the rest. ### The bar has risen What really changed is which links count. In December 2022 Google said its AI spam system, SpamBrain, now neutralises unnatural links. It spots sites that buy links and sites that exist to pass them on. Any credit those links passed is simply lost. Neutralise is the word to remember. The usual fate of a bad link today isn't a penalty. It's nothing. Google stops counting it and doesn't tell you. Sounds harmless, right up until you remember you paid for it. Picture ReelRadar's owner buying a cheap link package from sites that publish anything for a fee. A month later every link is live, the backlink tool's graph points up, and the rankings haven't moved. No penalty, no warning, no refund. It's the most common link building story there is. And buying links isn't a crime. Google's spam policy calls paid links a normal part of the web's ad economy, as long as they carry a sponsored or nofollow label. Google draws the line at paid links meant to pass ranking credit. So what makes a link count now? Next episode goes deep, but here's the short version from our field work. An indexed, independent page about a related topic, with the link inside the main text where a reader might actually click it. Drop any of those conditions and the link is worth less. Now our own position, plainly. Advanced off page methods still work in many industries. Consolidating power from strong, relevant old domains. Tiered campaigns. Carefully run private networks. We've watched them move rankings on real campaigns, and that's our field work. The other half is just as true. Most people who try these from a video lose money. They buy domains about the wrong topic, leave footprints that tie everything together, stack junk on junk, and quit after a month. Done well, these methods look like real authority. Done badly, they look exactly like what SpamBrain was built to catch. So we'll teach the why, never the recipe. ### Where AI answers look instead Now the newer question. More and more searches end in an AI answer. Do your links help you get into it? Mostly not directly. What the assistants seem to care about is whether people talk about you. Ahrefs ran this on seventy five thousand brands in December 2025. YouTube mentions and branded mentions across the web tracked how often a brand showed up in AI answers far more closely than anything else. Domain rating came in well behind. Raw backlink counts barely registered. Two details matter. Links that carry your brand name still did fairly well, which makes sense, because a branded link is also a mention. And the study only covered brands that already had real authority, so it says little about tiny sites. Our one caveat for the episode: this is correlation. Big brands get mentioned, linked and searched all at once. Muck Rack checked more than twenty five million links cited by ChatGPT, Claude and Gemini in May 2026. Eighty four percent came from earned media. Paid and advertorial content: zero point three percent. You can argue about how they sorted the categories. You can't argue that gap away. Links aren't out of the picture, though. An agency study ran ten thousand questions through GPT four o, and the best predictor of a brand getting mentioned was ranking on Google's first page. Backlinks on their own were weak. Put it together: links help you rank, and ranking helps you get mentioned. Real effect, just indirect. So if ReelRadar wants to turn up when someone asks an assistant where a new film is streaming, another batch of backlinks is the wrong project. Getting quoted by film journalists, discussed on forums and covered in videos is the right one. Ranking well in Google helps too, and that's where links come back in. Bing rankings tracked mentions too, a little less closely. One more reason not to ignore Bing, as we said in Part 1. Assistants that lean on a search index inherit that index's opinion of who deserves to rank. And before anyone starts a fleet of fake forum accounts: Google's guide to AI search says seeking inauthentic mentions isn't as helpful as it might seem, because its AI features run on the same spam systems as everything else. Seeded mentions are just the new paid link packages, with extra steps. Let's check it live. Same question in three assistants: where can I check UK streaming availability for new films? We log every source each one cites and sort them into big publications, forums or video, and small niche sites. Then we check whether the cited sites are the ones with the biggest link profiles. Whatever today's count shows, look for the pattern the studies describe. Sources people discuss, review and reference get cited. A site with a huge link profile and no reputation outside it often gets skipped. So earned coverage pays twice: a link for Google and a mention for the assistants. ### Does your niche need links? So, does your site actually need links? Ask four questions. Do the pages ranking for your main queries have strong link profiles? Are your competitors brands people search for by name? Do your buyers ask an AI assistant before they buy? And is your own page already the best answer? The first question takes ten minutes. Take your three most valuable queries and the top five results for each. Put them into any backlink tool and ignore the authority score. Count how many relevant sites link to each ranking page, and whether those links sit inside real articles. If the answer is almost none, those pages are winning on their own. Thin link profiles at the top? Content and technical work will take you most of the way, and links come later. Every result a known brand with years of coverage? Then you need links and mentions, of the kind that would exist even without search engines. That describes most competitive commercial niches. Most sites land somewhere in between. Take ReelRadar again. Its where to watch queries are owned by big entertainment sites with deep link profiles, so links are necessary there. Its questions about regional release dates are answered by small pages with almost no links. So links go to the money pages, and good content handles the rest. What does that kind of link look like? Take ReelRadar, our fictional film trailer and where to watch site. A film critic citing its regional availability data. A festival listing linking to its coverage. A news story quoting its numbers on how long films take to reach streaming. Each brings readers, a relevant anchor and a mention in one go. And here's a real one: a restaurant in Istanbul we work with, where technical fixes, content and off page work ran together. On the fourteenth of June, before our work began, it got twenty seven clicks from Google. On the eleventh of September, it got seventy. September averaged nearly forty eight a day. Visibility grew even faster. Daily impressions went from five hundred and twenty seven in late June, when measurement started, to one thousand nine hundred and forty five across September. That's up two hundred and sixty nine percent. The links did their job together with the content and the technical work, which is how real campaigns win. ### Wrap up The whole episode on one screen. Links are one input, named in the court record for relevance and quality, judged harder than ever, and only indirectly tied to AI answers. Next episode we go inside that input: how a link actually passes value, and why one can outweigh five hundred. The does my niche need links decision sheet is free on the episode page at irongoo.com. Fill it in for your three most valuable queries before you spend a cent on links.