"Just Make Great Content" Is Dead Advice. Here's What Actually Works Now.

You followed the advice. You hired good writers, published consistently, made the content genuinely useful, and your traffic went down anyway. "Just make great content" was never a strategy; it was a slogan, and the people repeating it stopped checking whether it still worked years ago. Quality is now the entry fee, not the prize. Here is what actually determines who gets seen.

The real problem: quality stopped being scarce

The "great content wins" advice rested on one assumption: that great content is rare enough to stand out. That assumption is dead. Generative tools can produce competent, accurate, well-structured articles on any topic in minutes, and millions of businesses are doing exactly that. When everyone's content is good, good is worth nothing. Even before AI flooded the market, Ahrefs studied roughly 14 billion pages and found that 96.55 percent of them received zero traffic from Google. The web has never rewarded quality by default. It rewards quality that something else makes visible.

Meanwhile, the reward for visibility itself shrank. Pew Research Center tracked the actual browsing behavior of 900 U.S. adults in March 2025 and found that when Google showed an AI summary, users clicked a traditional result only 8 percent of the time, versus 15 percent without one. Links inside the AI summaries got clicked on just 1 percent of those searches. Ahrefs measured it from the publisher side: an AI Overview on the results page correlated with a 34.5 percent lower clickthrough rate for the top-ranking page in April 2025, and by their December 2025 update the figure had reached 58 percent. Similarweb found zero-click searches for news rose from 56 percent to 69 percent in the year after AI Overviews launched.

So the math facing a business owner is brutal. The supply of good content exploded, the clicks paid out for good content collapsed, and the advice you keep hearing addresses neither fact. "Make great content" answers a question nobody is asking anymore. The real question is: when an AI or a search engine assembles an answer for your customer, why would it pick you?

Why most businesses get this wrong: they optimize the artifact, not the entity

The mechanism behind the failure is simple once you see it. Businesses treat content as the thing being judged. Search engines and AI systems increasingly judge the entity behind it: the brand, the author, the accumulated evidence that this source is real, consistent, and trusted by others.

The evidence for this is now quantified. Ahrefs analyzed 75,000 brands in 2025 to find out what predicts visibility in Google's AI Overviews. The single strongest factor was brand web mentions, at a correlation of 0.664. Backlinks, which most SEO budgets still chase, came in at 0.218. What other people say about you across the web predicts your AI visibility roughly three times better than the links you have built. Brands in the top quartile of web mentions earned up to ten times more AI Overview mentions than the quartile below them. The judging happens mostly off your website, and most businesses spend 100 percent of their content budget on the website.

There is a second mechanism most owners have never heard of: how machines select text. AI systems do not read your article the way a person does. They break it into chunks, convert those chunks into mathematical representations, and retrieve the chunks that best match the question. A 2,000-word essay where the answer emerges gradually across eight paragraphs is nearly useless to that process, no matter how well-written it is. A page where each section opens with a direct, self-contained answer gets extracted and reused. Mike King of iPullRank, twice named Search Marketer of the Year, calls the discipline of engineering for this "relevance engineering," and it treats content as something machines must be able to parse, score, and lift, not just something humans enjoy reading.

And the third mechanism: distribution was always doing the work that people credited to quality. The case studies behind "great content wins" almost always involved brands with existing audiences, email lists, communities, or PR muscle. The content was the visible part. The distribution was the engine. Strip the engine and copy only the content, which is what most small businesses did, and you get the 96.55 percent.

What the data actually shows works now

Four things have measurable, documented support. None of them is "publish more."

First, original information. Google holds a patent, US 11,157,557, on scoring documents by "information gain," meaning how much a page adds beyond what the user has already seen. The first major academic study of AI answer engines, the "GEO: Generative Engine Optimization" paper from researchers at Princeton, Georgia Tech, IIT Delhi, and the Allen Institute for AI, tested nine tactics across 10,000 queries and found that adding statistics, adding quotations, and citing sources each boosted a page's visibility in AI-generated answers by 30 to 40 percent. Fresh facts get cited. Restated facts get absorbed.

Second, off-site authority. The Ahrefs numbers above, mentions at 0.664 versus backlinks at 0.218, are the headline, but the detail matters: in the same research, YouTube mentions showed the strongest correlation of any single factor with AI visibility, around 0.737. Semrush's AI visibility research found that community and reference sources, Reddit and Wikipedia chief among them, consistently beat corporate websites as citation sources across industries. Your blog competes with the entire web. Your reputation on third-party platforms competes with almost nobody in your niche, because almost nobody in your niche is doing the work.

Third, being in the answer is not the same as being the recommendation. Semrush found ChatGPT cites sources in about 87 percent of responses but names brands in the text of the answer only about 21 percent of the time. Kevin Indig calls this the ghost citation problem: your page can be a footnote the user never opens while a competitor's name sits in the sentence the user actually reads. The goal has shifted from ranking to being named.

Fourth, rankings and citations have decoupled. Ahrefs found that only around 38 percent of AI Overview citations come from pages ranking in the top 10 organic results, which means most citations are being pulled from pages the old playbook would have called losers. Structure and specificity are getting content cited from positions that never used to matter. Rand Fishkin of SparkToro has spent years building the data case that all of this requires new scoreboards, and his conclusion is worth taking literally: in a zero-click world, traffic is a terrible goal, and influence on the buyer is the thing to optimize.

None of this says quality is irrelevant. Sloppy, wrong, or thin content disqualifies you. But quality has moved from being the strategy to being the prerequisite. The strategy is everything wrapped around it.

How to fix it: five moves that replace "make great content"

Each of these can be done by a non-technical owner or delegated in an afternoon. Together they form the actual playbook.

1. Build your entity, deliberately

Make it unambiguous to machines who you are. Use the exact same business name, founder name, and description everywhere: your website, LinkedIn, Google Business Profile, directories, podcast bios. Publish a thorough About page that names your people, credentials, and history. Have your developer, or a freelancer for a few hundred dollars, add Organization and Person structured data to your site. Put a real author with a real bio on every article. This sounds mundane. It is the foundation the Ahrefs correlations are measuring: a consistent entity the model can recognize and accumulate trust around.

2. Restructure your key pages for extraction

Take your ten most important pages and rewrite them so every section starts with the answer. Turn headings into the actual questions customers ask, and make the first sentence under each heading a complete, standalone answer with a specific fact in it. Keep statistics in text, not locked inside images. Add FAQ sections with real questions. You are not dumbing the content down; you are making it liftable, which is what the GEO research shows gets rewarded.

3. Give every piece a distribution plan before you write it

New rule: if you cannot list five places a piece will be pushed, do not write it. Minimum viable distribution for one article: a native LinkedIn post with the key finding, an email to your list, a tailored pitch to two industry newsletters or podcasts, and a genuinely helpful answer in one community thread where the question comes up. If you can turn the piece into a short video, do that too, because YouTube mentions are the strongest AI visibility signal Ahrefs measured. One well-distributed article beats five published-and-prayed ones, every time.

4. Put one piece of original information in everything

Before publishing anything, ask: what is in this that exists nowhere else? It can be small. A real client number, an observation from last month's projects, a survey of 50 customers, a documented test with the result you did not expect. This is the information gain principle made operational, and it is also what makes a piece quotable by the AI systems your customers are asking. If a piece has no original information, it is volume, not strategy.

5. Consolidate the commodity content you already have

Run a simple audit, or have your marketer run it: list every post, its traffic, and its purpose. Posts that say what a hundred other sites say, earn nothing, and serve no customer question should be merged into stronger pages or removed. A smaller site where every page has a reason to exist presents a cleaner, more authoritative entity than a sprawling archive of me-too posts. You will often see the surviving pages improve once the dead weight is gone.

What to measure, and when to expect results

If you keep grading content by pageviews, you will keep funding the wrong things. The clicks are structurally declining for everyone; Ahrefs calls the spreading gap between impressions and clicks the great decoupling. Falling traffic alongside rising impressions can mean you are increasingly present inside answers people read without clicking. So change the scoreboard.

Track these five. One: AI visibility, checked monthly by asking ChatGPT, Perplexity, Gemini, and Google AI Mode the ten questions a buyer would ask before choosing a provider like you, in a clean session, recording whether you are named, cited, or absent. Two: branded search volume, the cleanest signal that off-site work is landing, visible free in Google Search Console. Three: mentions, counted with a free alerts tool or an SEO platform, including podcasts and YouTube. Four: AI and referral traffic, small numbers but high intent, visible in your analytics. Five: the business numbers, inquiries, calls, and revenue attributed to organic and AI channels, because that is the point of all of it.

Timelines: restructured pages can show up in AI answers within 4 to 8 weeks because extraction improvements act fast. Entity and mention building is slower, typically 3 to 6 months before AI tools start naming you on niche queries, 6 to 12 months for competitive ones. The vanity traps to refuse: posts-per-month as a KPI, total traffic as the headline number, ranking reports for keywords whose results are fully answered above the fold, and word counts as a proxy for quality. Measure whether the machines name you and whether buyers contact you. Everything else is noise.

Frequently Asked Questions

Is SEO dead now that AI answers most searches?

No, but a specific version of it is dying: publishing keyword-targeted articles and collecting easy clicks. Search is still where buying decisions start, and the systems answering those searches still rely on sources they can trust and extract from. The work has shifted from ranking pages to building a recognizable entity, publishing original information, and earning mentions across the web. Businesses that make that shift are gaining visibility while their competitors watch traffic reports decline. The discipline is alive; the old playbook is not.

What is entity authority and why does it matter?

Entity authority is the accumulated evidence, across the whole web, that your business is a real, consistent, trusted source on a topic. It is built from consistent naming everywhere, real author identities, structured data, and above all third-party mentions on sites, podcasts, videos, and communities. It matters because the data shows machines lean on it: Ahrefs' study of 75,000 brands found web mentions predicted AI Overview visibility about three times better than backlinks. In practice, AI systems recommend brands the web already talks about. Your content is the artifact; the entity is what gets judged.

Should I delete old blog posts that get no traffic?

Audit first, then consolidate, and only delete what serves no purpose. Posts that restate generic information, attract no traffic, and answer no real customer question can be merged into a stronger page or removed, with redirects so you keep any link value. Keep anything with original data, first-hand experience, or steady conversions even if traffic is modest, because that is exactly the material AI systems cite. The goal is a tighter site where every page adds something unique. Most businesses find a smaller, sharper site outperforms the bloated archive it replaced.

"Just make great content" survived for a decade because it flattered everyone and obligated no one. The data has now called the bluff. Quality gets you to the starting line; entity authority, extractable structure, real distribution, and original information decide who gets named in the answer. Do those four things and your good content finally gets the visibility everyone promised it would earn on its own.