Why GEO Has No “Backlink Formula”: AI Search Is Redefining Credibility

The other day, I was reviewing recent changes in ChatGPT and Perplexity citation sources with a colleague who has been in SEO for many years.

We looked at several industry questions and casually compared AI citations for some brands under different query phrasings. Midway through, he scrolled back up a few times and suddenly asked me: “If channels like Reddit and industry media are increasingly appearing in citations, does that mean they’re the new backlinks?”

He paused, then added: “If we identify a few high-authority communities and media outlets and consistently create content for them, can we replicate GEO results?”

This question is similar to one that has recurred in the SEO industry over the past two decades: is there a relatively stable variable that, if consistently invested in, can yield fairly predictable search returns?

I didn’t dismiss his idea immediately.

Community content can indeed influence how AI accesses and understands information. Industry media, professional websites, and corporate sites can all become information sources within citation systems like ChatGPT citations and Perplexity citations. Continuously researching changes in these sources is, in itself, an important part of a GEO strategy.

But if we jump to the conclusion that Reddit is the new backlink, or that a few types of websites constitute a fixed formula for GEO, I think we’re oversimplifying things.

What truly deserves attention isn’t which channel suddenly becomes important, but how AI search is deciding whether a piece of information is worth referencing.

Why hasn’t AI search replicated Google’s backlink system?

I started doing SEO at a Hong Kong digital marketing company in 2005, and I’ve witnessed the transition from Yahoo to Google. Looking back over these two decades, search changes have never been just about a different-looking page; they’re about how search engines assess the value of information.

In the Yahoo era, businesses cared a lot about directories and entry points.

Back then, being listed in an important directory meant an additional opportunity to be discovered by users. Businesses thought: “Can I get into these entry points?”

In the Google era, that logic changed. Search engines began crawling web pages at scale, understanding them, and evaluating page value through multiple signals, including the link ecosystem. SEO gradually developed a mature system covering keywords, pages, technology, content, and links.

Links thus became one of the key signals that SEO practitioners studied for the long term.

Now, in the AI search era, the question has changed again.

When users ask questions in ChatGPT, Gemini, Google AI Mode, or Perplexity, they often aren’t waiting for ten blue links to click through one by one to find the answer. AI tries to understand the question, organize information, and generate a fairly complete response.

This means the competition behind AI search visibility is no longer just about “which webpage should rank higher.”

It also involves: what information is most relevant to the question, what sources best support the answer, whether different sources can corroborate each other, and whether brand information is clear enough that the model isn’t prone to misunderstanding.

So I’ve always believed that when discussing SEO vs GEO, we shouldn’t simply frame it as GEO replacing SEO.

SEO hasn’t disappeared.

What search evaluates is simply expanding.

Why do businesses always look for a “GEO backlink formula”?

I understand this mindset very well.

SEO has developed a fairly mature working language over time. SEO teams study backlink counts, domain quality, page ranking factors, and keyword difficulty, and use past data to decide where to allocate resources next.

There’s nothing wrong with that experience.

It has helped many businesses turn search from a vague marketing activity into a process that can be broken down, executed, and reviewed.

So when AI search emerged, it’s natural for people to look for new corresponding relationships: SEO tracks rankings, so what does GEO track? SEO has backlinks, so does GEO have some kind of “AI backlink”? If we used to study Domain Authority, should we now just study which domains are more likely to appear in AI citations?

This concern is valid. But the issue is that what AI search evaluates has already changed.

Traditional search largely involves ranking among a large set of web pages, while generative search also needs to accomplish something else: organizing different pieces of information into an answer.

This involves not just page evaluation, but also information synthesis.

And it involves validation across multiple sources.

So when studying AI search ranking factors or LLM visibility, if we only look for a single quantifiable “backlink substitute,” we can easily miss the more important question: how does AI actually understand your brand, and is there clear, consistent information across the internet to help it form that understanding?

Why is there no single universal source among Reddit, industry media, and corporate websites?

There’s been a lot of discussion recently about Reddit AI citations in the industry. I’ve also been tracking changes in citation sources across different platforms.

Communities are certainly worth attention.

• Community content provides real user discussions, experiences, and feedback, which is especially valuable for certain consumer decisions, product comparisons, and experiential questions.
• Industry media offers validation from outside the company itself, supplementing brand background, professional viewpoints, industry events, and product information.
• Corporate websites remain a key source of factual brand information, especially product specifications, service scope, company introductions, technical details, and official statements, which all need to be accurate, clear, and well-structured.

But I wouldn’t recommend treating any single one of these sources as the universal channel for GEO.

The reason is simple: AI citation sources change depending on the question, the platform, the content type, and the information environment.

The same brand may not be cited from the same sources on ChatGPT as it is on Perplexity; and the presentation on Gemini AI search versus Google AI Mode shouldn’t be assumed to be identical either.

Even on the same platform, different questions require different types of information.

When one user asks “which types of businesses is a certain software suitable for,” and another asks “what are the technical specifications of a product,” the kind of evidence AI needs is fundamentally different.

So instead of asking “which website is most likely to be cited by AI,” I’d rather start with two questions:

First, is your information easy enough to understand?

Second, beyond yourself, are there other sources that can reasonably corroborate this information?

These two questions come closer to the heart of brand authority in AI search than looking for a so-called universal channel.

Why does AI choose to trust certain content?

Strictly speaking, it’s hard to compress the complex internal evaluation processes of different AI platforms into a few fixed metrics. I also don’t endorse phrases like “cracking AI citation rules.”

But from practical GEO work, we can observe some directions worth tracking consistently.

For example, whether a piece of content clearly answers the user’s question; whether the author or website has relevant professional background; whether core facts about the same brand are consistent across its official website, industry media, and other public sources; whether important pages are consistently maintained; and whether the content structure is easy for search systems to retrieve and understand.

The real shift here is that businesses are moving from “channel competition” to “credibility competition.”

When my team and I proposed AIPO, one of the considerations was to break down this kind of work.

The Platform aspect isn’t about telling businesses there’s a fixed citation formula for a certain platform. It’s about observing the differences in information presentation and citations across platforms like ChatGPT, Gemini, Google AI Mode, and Perplexity.

The Content aspect focuses more on what the business says, and whether that content truly answers user questions. Content planning and platform publishing shouldn’t just chase publishing volume, but should consider topics, query phrasing, information structure, and relationships between different sources.

In other words, a GEO content strategy isn’t about copying the same article to more websites.

It’s more like building an information system that can be understood, retrieved, and cross-validated.

From the SEO backlink mindset to the GEO source mindset

If we look at the changes over the past two decades together, I’d describe the differences across three generations of search as follows:

This change doesn’t mean links are no longer important, nor does it mean traditional SEO experience in content, technology, and website building is obsolete.

On the contrary, many fundamental SEO capabilities are still the foundation of GEO.

The shift is that we can no longer understand AI search trust solely through “how many links a webpage has.” Businesses also need to start observing: whether brand information is consistent in different places, whether AI can accurately recognize the brand entity, whether relevant content can answer real questions, and whether external sources can provide reasonable information validation.

This is one of the biggest cognitive shifts I see between GEO optimization and traditional SEO.

How should businesses rethink GEO?

If a business is still approaching GEO the way it approached backlinks in the past, it will likely ask: “Where should I publish to get AI citations?”

Now I’d recommend asking a different question:

“Has my brand information already formed a verifiable knowledge network?”

These two questions look similar on the surface, but they imply completely different resource allocations.

The first approach leads teams to constantly chase channels: this month it’s Reddit, next month industry media, and when citations from a certain type of professional site increase, they quickly shift resources there.

The second approach focuses on longer-term information assets.

• Whether brand name, business scope, and product/service information remain broadly consistent across the official website and other public sources.
• Whether the website structure is clear enough, and whether important pages use structured data like Schema to help search systems understand them.
• Whether content truly answers target customers’ questions, rather than repeatedly producing similar articles around keywords.
• Whether the brand’s presence and citation sources on platforms like ChatGPT, Gemini, Google AI Mode, and Perplexity are changing.

Within the AIPO framework, this isn’t just a Content issue.

The website optimization aspect needs to address page structure, Schema, and information organization; content planning and platform publishing needs to build content that corresponds to real user questions; and performance monitoring needs to continuously compare changes across different AI platforms, different query phrasings, and different time periods.

See clearly first, then decide what to optimize.

I think this is more practical than first looking for a so-called universal GEO channel.

“If there’s no formula, then what exactly should businesses do for GEO?”

I hear this question often.

After listening to my explanation, the SEO colleague earlier fell silent for a moment, then asked: “If there’s no clear formula like backlinks, then how do businesses know where to spend their money on GEO?”

That statement is half right, half wrong.

The right part is that GEO is indeed harder to explain with fixed metrics than SEO was in its mature stage. Answers, citations, and information organization can all vary across different AI platforms, and measuring results requires new observation methods.

But having no fixed formula doesn’t mean there’s no optimization direction.

Google SEO itself has never been about a single ranking button. Technology, content, links, user intent, and site quality have always needed to work together over time. It’s just that after years of industry development, we’ve become accustomed to this language, so it’s easy to feel it was always inherently certain.

At the end of 2022, I began systematically researching AI marketing. I was already using Claude to observe changes in generative answers back then. At that time, I felt more and more strongly that we weren’t facing another traditional search box, but a new way of organizing information.

So GEO monitoring targets should change accordingly as well.

In which questions does the brand appear?

Which content is cited by AI?

Which key facts aren’t recognized, or are even distorted?

For the same question, what are the differences on ChatGPT, Gemini, Google AI Mode, and Perplexity?

These questions can all be continuously recorded, compared, and analyzed.

Not being able to compress GEO into a formula doesn’t mean GEO can’t be measured.

On the contrary, it requires businesses to build a monitoring system that’s closer to real AI search visibility than “how many backlinks were published.”

AI search competition is shifting from traffic competition to credibility competition

Looking back at three generations of search, I think the change is quite clear.

In the Yahoo era, we competed for entry points.

In the Google era, we competed for rankings.

Now in the AI search era, businesses also need to compete for another position: when AI answers a question related to an industry, product, or brand, can your information become part of what it references to understand that question?

In the AI search era, businesses aren’t just competing to be seen; they’re competing to become a source AI is willing to reference when understanding the world.

This is also why I don’t believe GEO will eventually converge into an eternal “backlink formula.”

Platforms will change, citation sources will change, and user query phrasing will change.

But businesses can build some relatively long-term assets: clear brand facts, consistently updated professional content, well-structured websites, and information assets that can be validated across multiple public sources.

SEO and GEO are not an either-or choice.

From Yahoo to Google, and now to AI search, each search paradigm shift hasn’t completely discarded the previous generation’s capabilities. More often, businesses need to add a layer of capability adapted to the new information environment on top of their existing search assets.

If you’re an SEO manager, I’d suggest first re-evaluating your existing content assets. Don’t just look at keyword rankings and organic traffic; also start observing whether important content is entering AI search answers and citations, and whether AI’s understanding of your brand is accurate.

If you’re responsible for digital strategy, I’d suggest evaluating GEO as part of your long-term search strategy. First understand how your brand is presented and cited across different AI platforms, and how you compare to competitors, then decide what needs to be done for your website, content, and external sources respectively.

Related Questions

Does GEO have a fixed formula similar to SEO backlinks?

Not currently. AI search citation and answer mechanisms are still evolving. Different platforms and different questions may focus on different information sources and content characteristics, so continuous monitoring is more appropriate than applying a single formula.

Is Reddit the most important channel for GEO?

No. Community content like Reddit may be one of the sources AI uses to gather user discussions and experiential information, but businesses shouldn’t rely on a single channel to build AI search trust. Corporate websites, industry media, professional content, and other public sources all have their respective roles.

Why does AI search cite a particular website?

From practical observation, it’s usually worth paying attention to content relevance, information clarity, professional background, information consistency, and whether the content is easy to retrieve and understand. However, mechanisms differ across AI platforms, so it’s not appropriate to treat these observations as fixed AI citation formulas.

Is a corporate website still valuable in the GEO era?

Yes. A corporate website remains an important source for expressing brand facts, product information, service scope, and professional content. In the GEO stage, it becomes even more important to focus on site structure, Schema, information accuracy, and whether content clearly answers user questions.

What is the biggest difference between GEO and SEO?

SEO focuses more on page discovery, understanding, and ranking within search results. GEO further focuses on how AI understands, presents, and cites business information when generating answers. The two are additive, not substitutive.

How should a business start with GEO?

I’d recommend first checking your brand’s basic information and building a set of questions relevant to your actual business, then observing how platforms like ChatGPT, Gemini, Google AI Mode, and Perplexity currently present your brand. Once you identify information gaps, misunderstandings, and citation differences, decide how to adjust your content, website structure, and performance monitoring accordingly.