Is SEO dead now that AI Overviews are everywhere?

No, but the mechanism changed. The data above shows organic clicks falling even when rankings hold steady, which means visibility and traffic aren't the same metric anymore. SEO still decides whether you show up at all — in a blue link, an AI Overview citation, or a local pack — it just no longer guarantees a click the way it did five years ago.

Do backlinks still matter for ranking?

Yes, based on the correlational data above from Backlinko and Ahrefs — link authority and referring domain count are still among the strongest measurable correlates of first-page rankings. What's changed is that backlinks alone, without real expertise and useful content behind them, are a much weaker bet than they were a decade ago.

What's the difference between SEO and GEO (generative engine optimization)?

SEO is optimizing to rank in a results list and earn a click. GEO, covered above, is optimizing to be the sentence an AI system quotes or paraphrases inside its answer, which may or may not include a visible link. They share a foundation — authority, clarity, real expertise — but increasingly reward different specific tactics, per the industry research cited above.

Should I write new content or update what I already have?

Both, but most businesses are lopsided toward only the first one. The freshness data above suggests updating existing, previously-performing pages is often faster and cheaper than starting from zero, because those pages already carry indexing history and, often, backlinks new pages don't have yet.

Is E-E-A-T an actual Google ranking factor I can optimize for directly?

Not directly — Google is explicit that E-E-A-T is a framework for human quality raters, not a ranking input itself, as covered above. But it describes, in plain language, the kind of content Google's automated systems are tuned to reward, which makes it a genuinely useful mirror to hold up to your own site even though there's no literal E-E-A-T score to chase.

SEO Isn't Dead, But 'Rank #1' Stopped Being the Win: What Actually Works Now

The data behind why ranking first doesn't mean what it used to, what Google's own quality raters actually look for, and what businesses should be doing instead of chasing yesterday's SEO playbook.

Somewhere in the last two years, the promise that made SEO make sense to a small business owner quietly stopped being true. The promise was simple: rank on page one, and the clicks follow. Spend the time, do the work, earn the position, get the traffic. That was always the deal, and for a long time it held.

It doesn't hold the way it used to. Not because Google stopped ranking pages — it still does, the same mechanics still apply — but because ranking well and getting clicked are no longer the same outcome. A business can hit position one for a keyword it's been chasing for a year and see less traffic from it than it would have gotten from position three, two years ago. That's not a hypothetical. It's the actual, measured shape of search right now, and it's worth understanding in some detail before deciding what to do about it.

The click collapse, and why it's not a rumor

The headline number is blunt enough that it's worth stating plainly before getting into why it happened. A randomized field experiment covered by Search Engine Journal found that Google AI Overviews reduced organic clicks on triggered queries by 38%, while user experience ratings stayed unchanged. That's not a correlation that could be explained away by seasonality or a bad sample — it's a controlled study specifically isolating the effect of the AI summary sitting above the results.

The same reporting pulls in two other measurements that make the picture harder to argue with. Pew Research found users click 8% of the time with AI Overviews, compared to 15% without — roughly half the click rate, measured from real search behavior rather than estimated from keyword tools. And separately, Ahrefs analyzed GSC data and reported a 58% drop in click-through rate for top-ranking pages when AI Overviews appeared.

It's not confined to queries where an AI Overview actually shows up, either, which is the part that surprises most people who haven't looked closely at their own Search Console data. According to marketing agency Seer Interactive's analysis, reported by Search Engine Land, organic click-through rates for informational queries featuring Google AI Overviews fell 61% since mid-2024, while paid CTRs on those same queries plunged 68%; even on queries without AI Overviews, organic CTRs fell 41%. Put together, that means the decline isn't really about AI Overviews specifically — it's about a broader shift in how people use a results page at all, with AI Overviews as the most visible symptom rather than the sole cause.

Zoomed out further, the numbers describe something closer to a structural change than a seasonal dip. One analysis compiling research from multiple publisher and industry reports found that 60% of Google searches now end without any click to a website, up from 58% in 2024. A search that used to end with a click to somebody's site now, more than half the time, ends on the results page itself.

None of this means SEO stopped working. It means the thing SEO was actually buying you — a click — is now a smaller and less reliable slice of the outcome, even when the ranking work is done correctly. That's a real shift in what success looks like, and it changes what's worth optimizing for.

Why "get to #1" was always an incomplete goal

Even setting AI Overviews aside entirely, the idea that ranking first is the finish line was always a simplification, and the data on what actually correlates with high rankings tells a more specific story than "be the best page."

The most cited large-scale study on this remains the joint analysis Backlinko ran with Ahrefs across 11.8 million search results. It found that the #1 result in Google has 3.8x more backlinks than positions #2-10, according to analysis of 11.8 million Google search results. That's a genuinely strong signal — link authority correlating that cleanly with position isn't noise. The same study found that the average word count of a first-page result in Google was found to be 1,447 words, which is less a magic number to hit and more a sign that thin content rarely competes at the top regardless of how well-optimized it is technically.

What the same research quietly overturned is the assumption that page speed is a dominant ranking lever — a claim that gets repeated constantly in SEO sales conversations. The study's authors found that despite page speed being a confirmed Google ranking factor, fast-loading websites don't appear to have any ranking advantage over websites that load slowly, and site speed may be a minor ranking signal that doesn't translate into actual ranking improvements. That doesn't mean speed doesn't matter — it clearly matters enormously for conversion and user experience, which is its own separate argument — but it means speed isn't the lever that gets you to page one, and treating it as an SEO tactic rather than a trust and conversion tactic is a category error worth correcting.

Backlinks, on the other hand, keep showing up as one of the few things that correlates this consistently. A separate Ahrefs analysis found that pages with at least one backlink perform 10x better in organic search results than pages without backlinks, and that same research identified a clear correlation (r=0.38) between the number of referring domains and rankings, the strongest correlation of all examined ranking factors. So the old advice — earn real links from real sites, don't just publish and hope — hasn't aged out. It's just no longer sufficient on its own, because ranking well and being clicked have partially decoupled, per everything above.

What Google's quality raters are actually looking for

A lot of SEO advice throws around the term E-E-A-T — experience, expertise, authoritativeness, trustworthiness — as though it's a checkbox Google's algorithm tallies directly. It isn't, and Google has been unusually direct about that distinction, which makes it worth quoting precisely rather than paraphrasing loosely.

Google's own Search Central documentation states plainly that search raters have no control over how pages rank, and rater data is not used directly in our ranking algorithms. The company describes the relationship with a specific analogy worth repeating because it clarifies the whole system in one line: "we use them as a restaurant might get feedback cards from diners. The feedback helps us know if our systems seem to be working."

So E-E-A-T isn't a lever you pull. It's a description of what Google's human evaluators are trained to notice when judging whether the algorithm's output looks right — which makes it less a ranking factor and more a mirror. Google's blog post announcing the update explained the addition of the second "E," for Experience, this way: the update emphasizes content created from first-hand experiences, such as produc[t use, a real visit, or actual hands-on involvement in the topic] rather than content assembled secondhand from research alone.

Practically, that means the honest version of "optimizing for E-E-A-T" isn't a technical task at all. It's making sure the person or business behind a page is visible and real — a named author, a described process, evidence of actual work done — rather than trying to reverse-engineer a score that, by Google's own account, doesn't feed directly into the algorithm. The instruction that matters most here isn't "add credentials to your bio." It's "make it obvious a real person or business with real experience wrote this, because that's what raters are trained to look for, and rater feedback is what shapes the systems that eventually do decide your ranking."

The old lever that still works: fixing what you already published

If there's a genuinely underused opportunity hiding in all of this, it's not a new tactic — it's an old one that most businesses skip because publishing something new feels more productive than revisiting something old. The data doesn't support that instinct.

A widely cited HubSpot study found that updating old blog posts can increase traffic by up to 106%. That's not a marginal bump from fixing a typo — it's evidence that a page's age, relative to how current its information looks, is doing real work against it in search, independent of the quality of the original writing.

The mechanism behind that number gets illustrated well by a specific, documented case from Ahrefs. When the company's own content lead made mostly small edits to an old post, including mostly spelling and grammar errors, he increased traffic to that page by 486% — though it's worth noting, as the same source does, that the edit also included re-optimizing for a more relevant keyword and refreshing the publish date, not spelling fixes alone. The lesson isn't "typos are costing you rankings." It's that a page Google perceives as stale gets treated differently than the identical page perceived as current, and the perception can shift with a surprisingly small amount of real work.

This matters more, not less, in the AI Overview era described above, because a page that looks outdated is also less likely to be the one an AI system chooses to summarize or cite when it's deciding which source represents the current, reliable answer. Freshness isn't cosmetic. It's a signal that touches both halves of the modern problem — traditional ranking and AI citation — at the same time.

The new lever nobody had five years ago: getting cited inside the answer

If ranking #1 buys a shrinking click rate, the obvious next question is what actually replaces it as a goal — and an entire new discipline has formed around exactly that question. It goes by a few overlapping names — GEO, AEO, LLMO — but the practice is consistent: optimizing content not to be clicked, but to be the sentence an AI system quotes, paraphrases, or cites when it answers someone's question directly.

There's a real commercial case for taking this seriously rather than dismissing it as a fad. One analysis found that brands that are successfully cited in AI Overviews are seeing a 35% increase in organic CTR compared to those that are not — meaning the businesses that do get cited aren't just getting a consolation prize, they're outperforming the traditional click model, not just surviving inside a shrunken version of it.

Getting cited isn't disconnected from traditional SEO, either, which is a useful reassurance for anyone worried this means starting over. Search Engine Land's coverage of the practice notes that AI Overviews cite sources that overlap heavily with Google's top 10, and Perplexity's citations overlap about 60% with Google's top 10, too — meaning if you don't rank reasonably well in the first place, your odds of being cited by an AI system drop along with it. Ranking well is still the entry ticket. It's just no longer the whole show.

What has changed is which specific tactics matter once you're in contention. Some research suggests the overlap between traditional SEO winners and AI-citation winners is shrinking quickly — one industry summary put it starkly: the two now share fewer than 20% of their winners, down from about 70% two years ago, according to industry research summarized by Search Engine Land. Practically, the tactics that seem to matter for citation specifically include writing a clear, self-contained answer near the top of a page rather than burying the direct answer under three paragraphs of preamble, using real structure — tables, defined terms, labeled sections — that a model can parse cleanly, and citing your own sources transparently enough that a model treats your page as a credible node to quote rather than a page to skip past in favor of the source you cited.

What this actually looks like on a real page

None of this requires abandoning good writing for some new robotic format. It mostly requires discipline that good writing already rewards: answer the actual question early and plainly, back specific claims with specific, checkable sources rather than vague authority claims, and structure a page so both a human skimming on a phone and a model parsing the HTML can find the answer without digging. The overlap between "written well for a human" and "written well for an AI system" is larger than the SEO industry's sudden rebranding around GEO might suggest — it's mostly the same discipline, applied with slightly more awareness of how it gets parsed.

Where this actually leaves a small or premium business

None of the above is an argument that SEO is pointless now — quite the opposite. It's an argument that the specific target businesses have been aiming at for a decade has moved, and aiming at the old target with the old confidence is how you end up doing real work for shrinking returns. A few things follow directly from the research above.

Traffic is no longer the only honest scoreboard. If a page can rank well and still lose clicks to an AI summary sitting above it, judging an SEO effort purely by session count in Google Analytics is measuring an incomplete picture. Visibility — showing up in the summary, being the named source, appearing in the local pack — increasingly matters alongside raw click volume, not instead of it.

Real expertise stops being optional decoration. Because E-E-A-T describes what human raters are trained to reward, and because AI systems appear to favor sources that read as genuinely authoritative rather than generic, the businesses that document actual expertise — named people, real process, specific results — have a structural advantage over ones publishing generic, anonymous content, independent of any particular algorithm update.

Old content is an asset sitting on the table. Given the freshness data above, most businesses with even a modest content history have pages worth revisiting before they'd get equivalent value writing something brand new. A content audit — what's ranking but declining, what's outdated, what could be re-optimized for a sharper keyword — is often the highest-leverage hour a small business can spend on SEO this quarter.

Backlinks still matter, but from real, relevant sources, not volume for its own sake. The correlational data holds up. What doesn't hold up anymore is any strategy built on link volume alone without the underlying content being genuinely useful — that gap is exactly where AI-era scrutiny and old-fashioned algorithmic judgment increasingly agree.

The part that hasn't changed at all

It's worth ending on what didn't move, because it's easy to read a piece like this and conclude the ground shifted completely. It didn't. Google — and increasingly, the AI systems built on top of search data — is still trying to do the same thing it's always tried to do: surface the source that actually, verifiably knows what it's talking about, and put it in front of the person asking. Every mechanism described above, from E-E-A-T to backlink correlation to citation selection in AI Overviews, is a variation on that same underlying test.

What changed is the reward for passing that test. It used to be a click, reliably, almost every time. Now it's a click sometimes, a citation sometimes, and a mention inside an AI answer sometimes — and a business that's only built to capture the first of those three outcomes is leaving the other two on the table. The work required to earn all three is largely the same work it's always been: be real, be specific, be current, and be genuinely worth citing. The distribution of the reward changed. The requirements to earn it didn't.