Why GEO is not the panacea in the era of AI-powered searches

Influencing brand visibility in the responses provided by AI systems is something that often escapes GEO.

Internet searches have undergone numerous changes in just a few years. Approximately five years ago, these searches, once monopolized by Google, migrated (at least partially) to large marketplaces like Amazon. Two years later, social media became the primary platform for a significant portion of online searches, and now AI has burst onto the scene like a bull in a china shop.

While the customer journey for many internet users previously began with Google, ChatGPT, Gemini, and other AI systems have now become the entry point for many consumers. Indeed, a recent Criteo report concludes that large language models (LLMs) are the preferred entry point to the online customer journey for one in seven consumers.

This and other figures have raised concerns among many brands, who fear that if they don’t make their mark in the responses provided by AI, they could eventually become irrelevant. To address this fear, new disciplines like GEO (Generative Engine Optimization) have emerged, an evolution of traditional SEO that emphasizes brand visibility in the new AI era.

Is GEO a lifeline for brands in the AI ​​age?

However, influencing brand visibility in the responses provided by ChatGPT and similar platforms is by no means easy, and in some cases, efforts in this direction can be completely futile. This is due to the multitude of sources that LLMs draw upon when answering user questions.

Some studies highlight, for example, that AI systems favor premium media outlets when providing users with answers to their questions. And for brands, establishing an editorial presence in these types of media presents numerous challenges.

According to a report undertaken last year by Profound, ChatGPT, Gemini, and similar platforms predominantly favor reputable media outlets such as Reuters, Forbes, and The New York Post to answer user questions. However, premium media outlets are not the only source of information for LLMs (Learning Management Engineers), who also heavily rely on platforms like Wikipedia, LinkedIn, and Reddit. This is definitely not good news for brands, for whom gaining traction in Reddit threads, for example, becomes a near-impossible mission, given that the platform offers very little control over the content disseminated there. On LinkedIn, where much of the content shared on this platform relies on AI, brands also face significant challenges in influencing that content. Brands have little to no control over whether content is considered trustworthy or untrustworthy on Microsoft’s social network.

As for Wikipedia, a presence on this platform is similarly a treacherous territory for brands. If brands request changes to a Wikipedia entry (because they deem it appropriate and/or beneficial), the change may be viewed critically. And if the data contained in an entry of the famous online encyclopedia is incorrect, Wikipedians have no qualms about replicating it, which can potentially put brands directly affected by those errors in a difficult position.

“The image a brand projects in AI-generated responses can only be controlled to a certain extent, particularly on platforms where third-party content predominates, such as reviews and forums,” Jens Fauldrath of SEO consultancy get:traction told Horizont. It’s also worth noting that many specialized media outlets actively block LLM crawlers from accessing their archives, and that many platforms falling into the “dark social” category (messaging apps or email) are virtually invisible to AI systems.

Brands’ influence on AI-generated responses is inevitably limited.

«News articles, reviews, forums, Threads, and product comparison websites feed into almost all AI-generated responses, and these can only be influenced indirectly,» insists Fauldrath.

According to Olaf Pleines, Head of SEO at the digital agency Construktiv, GEO strategies primarily run up against two limitations. The first concerns the crawlability and readability of content by LLMs (Learning Media Managers). Platforms that are completely closed to LLM crawlers due to technical obstacles have no impact on the knowledge acquired by AI. The second limitation relates to content curation. «It’s not possible to place arbitrary messages on every single channel available to brands.» Wikipedia operates under very strict rules, large online forums have administrators who closely monitor every activity, and media editorial teams also actively monitor content,” Pleines points out.

When answering user questions, LLMs also grapple with technical peculiarities such as “citation drift.” This occurs when an AI system is presented with the same question twice in a row, potentially producing different (and in the worst-case scenario, diametrically opposed) answers.

Although it has become ubiquitous in many companies, GEO is still in its infancy and, unfortunately, cannot influence every single source consulted by AI. Furthermore, the way LLMs gather information is hampered by more questions than answers.

The main lever that brands can use to their advantage to try to influence AI is their own media (primarily the content hosted on their website). Secondly, brands should try to extend their reach into media, video, and multimedia content (YouTube, for example, is a source frequently cited by AI). However, in the other sources used by AI, brand influence is inevitably limited. And that’s something brands must consider when embracing GEO.

Source: www.marketingdirecto.com