GEO instead of SEO?

AI has radically changed search behaviour. Instead of comparing Google results, users receive a curated answer straight away. These answers usually include well-founded recommendations for providers, tools or solutions. This means that if your company is mentioned or cited, you immediately gain a trust advantage. Conversely, if your competitor is mentioned – and you are not – you lose visibility and, in the worst-case scenario, valuable enquiries or even contracts. In many sectors, therefore, it is precisely at this point that future digital visibility – and thus competitiveness – is determined.

You hear this statement a lot at the moment, but it’s not quite as simple as that. We need to take a more nuanced view here. The fact is that companies which have always operated according to genuine quality principles also benefit when it comes to AI-powered search. First and foremost, those that have structured their content in line with the E-E-A-T principle, as defined by Google.

• Expertise
• Experience
• Authority
• Trust

This content is exactly what AI systems also favour. Language models aim to deliver reliable, helpful and trustworthy information. 

For years, many websites have produced content that, whilst optimised for search engines, was not really written for people. As before – or perhaps more than ever – the motto is ‘USER FIRST’. We write for people, not machines. Superficial texts, generic content and interchangeable content are becoming less and less effective. We are currently receiving many enquiries regarding drops in rankings, as the content no longer meets Google’s quality criteria. This is because Google, too, has raised these standards immensely. So anyone who optimises their content for AI search will also benefit from better Google rankings. These remain relevant, but the balance will increasingly shift towards the importance of ‘rankings in AI systems’.

Our key findings are: AI loves content that isn’t interchangeable. What works:

• genuine expertise
• clear stances
• specific examples
• case studies
• personal experiences
• Comprehensible results

What doesn’t work:

• Generic AI content
• Smooth marketing copy
• convenient copy-and-paste

Companies must demonstrate:

• what they are truly capable of
• what specific problems they solve
• who their solution is particularly relevant for

And they must do so in a very personal way that genuinely helps.

Language models aim to use the most reliable sources possible. This means, for example:

• genuine client projects
• verifiable results
• studies or our own data
• credible references
• Ratings and reviews

For example, if a company has a lot of good reviews, this should not only be visible on Google – it should also be clearly highlighted on its own website.

The key strategic question today is: How do I create content that is so helpful, accurate and credible that language models favour it?

That’s exactly what we’ll be discussing in the next post. In it, I’ll show you specifically:

• which content structures work particularly well
• how content should be structured
• and how businesses can increase their chances of being recommended by AI systems.

Amidst all the enthusiasm for new technologies, there is one thing we must not forget: AI is a tool – not a substitute for critical thinking.For me , this is wherea clear line is drawn. AI can be of enormous help in structuring content, developing ideas or processing complex topics more quickly. But it must never lead us to accept content without scrutiny and, in doing so, to outsource our own thinking as well.

When blindly trusting AI becomes rather embarrassing – striking examples

In the case of Mata v.^Avianca¹, a lawyer submitted pleadings in which ChatGPT had simply provided fabricated precedents. The error was uncovered, the court imposed sanctions – and the case became a rather embarrassing lesson in what happens when you blindly trust AI. And anyone who now thinks this was an early slip-up from the early days is, unfortunately, mistaken. An international^study² by the European Broadcasting Union from 2025, in which the BBC, ARD and ZDF, amongst others, were involved, reaches a sobering conclusion: 45 per cent of all AI responses contained at least one significant error. 31 per cent showed serious problems with source citations. 20 per cent contained clear inaccuracies, hallucinations or outdated information.

Artificial dumbing down?

For me, this is the crucial point: AI can be very helpful, but you certainly shouldn’t trust it blindly. Anyone who adopts content without checking it risks errors that are not only embarrassing but, in the worst case, can also bedamaging to business. Quality does not come from convenient copy-and-paste, but from verification, contextualisation and human responsibility. In this context, some are already speaking of a kind of ‘artificial dumbing down’, where information is simply adopted without being properly scrutinised.

My request:

AI should be used to improve quality – not to diminish it. Especially at a time when content can be produced ever more quickly, trust, quality, personality and originality are becoming the new currency. For the real danger lies not only in errors, but also in a slick,uniform AI-generated mishmash where, in the end, everything sounds the same and the essence is lost: genuine opinion, genuine experience and genuine recognisability. Ultimately, therefore, it is not the fastest-produced content that wins out, but the most credible and original. And that is precisely what is crucial today for being perceived as relevant and recommended by AI systems.