Generative AI is reshaping Google search results, giving brands one of the biggest headaches since the introduction of SEO. As more people turn to tools like ChatGPT, Claude, and Google AI Overviews – which deliver direct answers rather than lists of links  – visibility is only the first hurdle. The real challenge, central to Generative Engine Optimisation (GEO), is ensuring brands are represented accurately and not undermined by hallucinated or incomplete information.

Jamy Wehmeyer is researching how brands can safeguard authority and trust inside a rapidly changing algorithm that nobody fully understands. He is the co-founder and CEO of Asky, a Stockholm-based platform for GEO that helps brands improve their visibility in AI‑generated search results. In conversation with Aspidistra, Jamy sat down to answer questions on visibility, trust and the future of AI search.

Why AI is redefining visibility in search

As search shifts from lists of links to direct answers inside AI tools, what’s changing most about how brands gain visibility before a click ever happens?

The previous strategy of ranking highly on Google, enticing users with a good title and meta description, and then converting them on your website, no longer really works. You lose that level of control over what users see as soon as they dig deeper.

Instead of just searching “I need a new payroll provider for HR,” users now ask ChatGPT and get a list of recommendations. The first challenge is simply showing up in that list. The next challenge is that visibility in AI answers is incredibly hard to observe directly. Unlike traditional search, where you can look at rankings and traffic, AI‑generated answers are non‑deterministic and personalised. What one user sees isn’t necessarily what another user sees.

Because of that, asking a single chatbot a single question doesn’t tell you much. You may appear in one response and be absent in the next, even if nothing has changed. Organisations often mistake these one‑off checks for insight, when in reality they’re just anecdotes. To understand visibility at all, you need to analyse patterns across many prompts, over time, and from multiple perspectives.

Why traditional SEO and measurement models are breaking

Once organisations realise that, what makes measuring, testing and influencing AI visibility so difficult in practice?

The main issue is that meaningful testing has to happen at scale. You need to ask thousands of variations of the same question to understand whether answers converge toward a stable distribution or are simply random. Studies have shown that answers do stabilise over time, but only if you’re running enough queries to see the pattern.

On top of that, personas matter. A 40‑year‑old technical decision‑maker at a multinational company will often get different answers than a teenage e‑commerce founder looking at the same problem. These implicit profiles influence how information is synthesised, which means brands need to understand how they appear across their different ICPs (Ideal Customer Profiles).

Internal testing makes this even harder. When a CEO or marketer asks about their own company, the chatbot may already have contextual bias from previous interactions. That makes it almost impossible to know what a neutral third party would see.

With clicks no longer the main KPI, what should brands focus on when AI answers become the first and sometimes only point of contact?

Currently, the best recommendation is to look at how much you’re investing in AI visibility, what your expected performance would have been without it, and how actual revenue compares. It’s not precise, but until better attribution exists, it’s the most realistic way to evaluate impact.

For example, you can tie AI visibility to something measurable, like how often you show up in AI-generated answers, and whether that is improving over time. You can also measure how often your brand is mentioned, and when it is, is the information accurate? Are the claims being made about your product correct?

What’s really broken is the classic customer-journey tracking model. You no longer see a clean path from Google search to website visit to purchase. A user might discover you entirely inside ChatGPT, go through their whole decision process there, and then land directly on your pricing page and buy. From your analytics perspective, you have no idea what influenced the decision, what messaging resonated or what feature convinced them.

At that point, measuring AI visibility looks a lot like brand marketing: similar to billboards, there’s no direct attribution, so you rely on proxies like revenue trends, forecasts, and before‑and‑after performance.

The leadership and strategy reset

What mindset shift should leaders make as they move into this new phase?

The basics of SEO are still essential. What AI adds is a new layer: understanding. AI systems don’t just scan pages anymore; they read the text, interpret meaning and reason about what they see. That means content now needs to be written in a way that’s easy for AI to cite, reference, and confidently mention.

This is where the mindset shift happens. It’s no longer enough to have a single well-optimised page. AI synthesises information from multiple sources to produce one recommendation. If, for example, one blog post says you specialise in the Nordic market, and another says you specialise in the Mediterranean market, an AI system may read both and become unsure about your positioning. And when the AI isn’t confident, it’s less likely to recommend you at all.

This extends beyond what you publish yourself. If third-party platforms describe your brand differently from how you position yourself, that inconsistency becomes part of what the AI weighs.

The uncomfortable truth is that AI can be naïve, making clarity, consistency and authority absolutely critical. The leadership mindset shift is moving from “we’re optimised” to “are we clearly and consistently understood everywhere AI might look?”

What is GEO, and how is it different from traditional SEO?

There’s a common misunderstanding that GEO is the new SEO, and that it’s one or the other. In reality, it’s a combination of both, but it requires its own strategy. It’s not about repurposing your SEO budget and calling it GEO; it demands different tactics.

This isn’t just about structuring a blog post differently because “that’s how we did SEO.” It requires a holistic approach. Not only what you do on your website or in your advertising, but also how you appear on third-party platforms. How do other people talk about you? Is your message consistent everywhere?

Tracking also becomes critical. What has a competitor written about you in a listicle? Is it accurate? Do they say, for example, that you don’t offer keyword data? If that’s incorrect and ChatGPT reads and believes their listicle over what you say yourself, that’s still a negative signal. Having a holistic overview and managing that entire ecosystem is probably the main difference from traditional SEO.

Playing the long game in generative search

Given how quickly AI systems evolve, how should companies think about balancing short‑term wins with long‑term brand health?

Some strategies might work today but won’t work two weeks from now, and some might even get you punished if you keep using them. We’ve already seen this happen. One competitor of ours jumped up the rankings by publishing content like “AI visibility in [city].” They took one article and duplicated it word-for-word for 24 different cities, only changing a few variables like city and country names. It worked really well for a short period.

But now they’re being punished. Even though those articles are no longer live, their regular content doesn’t perform as well anymore because they’ve picked up a negative signal. That’s the tricky part right now: balancing what might work in the short term versus what’s actually healthy for your brand in the long term.

SEO history has already taught us this lesson: trying to game the system usually ends up hurting you more than it helps.

How should companies structure their content differently so it works well for AI systems without sacrificing clarity or trust for human readers?

A common approach is structuring content very explicitly: using questions as paragraph headers and answering them directly in the first one or two sentences. It makes it extremely easy for AI systems parsing your site to recognise that the question is semantically similar to what a user is asking. AI doesn’t want to read a lot to find an answer – that creates a shift in how we write.

Traditionally, we’ve written content to feel engaging and human, with variation and storytelling. But from an AI perspective, “boring” is often good. It means the model doesn’t have to think hard to extract meaning.

Because of this, some companies are experimenting with separate content strategies. For example, having a highly structured, AI-friendly knowledge or blog section that’s optimised for citation and comprehension, while keeping more visual, narrative-driven content for human readers. The AI-focused content is still written as content, but it’s structured so the AI can understand, summarise and reference what you do almost instantly.

Local signals, media and the new opportunity curve

How do language and location affect how AI engines search for and surface information, especially in local markets like the Nordics?

We ran a study looking at what happens when the same question is asked in Stockholm but in different languages. Does AI prefer Swedish sources because of location, or English sources because of language? That led us to insights around query fan-out.

AI doesn’t just search for the exact question. It breaks it into multiple related queries like “best payroll provider” or “payroll providers in Stockholm,” then evaluates those results to decide what to trust.

Language plays a critical role. Both location and language influence fan-out queries, and in the Nordics, the sources surfaced differ from more US-centric search behaviour.

One key takeaway is that English-only content on a Swedish site isn’t enough. To show up locally, you need Swedish content, but when questions are asked in Swedish, roughly up to half of the fan-out still happens in English. That means you need both.

Do video and other media formats also play an important role in AI search and answers?

Zooming out, this comes down to owned and earned content. Earned content now includes YouTube, TikTok, Instagram, Facebook, Reddit and other community-driven formats.

AI systems don’t rely purely on traditional search rankings. They use their own indexing, follow links from sources they already trust, and analyse where users talk about brands organically. This is where the real opportunity lies: AI doesn’t depend on classic signals like popularity or follower counts to decide what’s credible.

YouTube is a good example. AI often cites videos from channels with only a few hundred subscribers or minimal views. Instead of judging reach, it checks whether the information aligns with what it sees across other sources. If it does, the video can still be cited as evidence.

The same applies across social platforms and forums. AI doesn’t just trust what you say about yourself. Instead, it looks for confirmation from unrelated third parties. When your message is consistent across external sources, the AI has more confidence in surfacing it.

In this model, credibility is less about brand size and more about consistency and clarity. For smaller companies that genuinely know their space, that creates a real opportunity.

Frequently Asked Questions about Generative Engine Optimisation (GEO)

What is Generative Engine Optimisation (GEO)?

Generative Engine Optimisation (GEO) is the practice of improving how brands are understood, cited and recommended in AI-generated search answers, such as those from ChatGPT or Google’s AI Overviews.

How is GEO different from SEO?

SEO focuses on ranking web pages in search results, while GEO focuses on ensuring accurate, consistent representation inside AI-generated answers before a user ever clicks a link.

Why is AI search a brand risk?

AI tools can omit brands entirely or present incorrect information. If an AI system lacks confidence or clarity about a brand, it may not recommend it at all.

How can brands improve AI visibility?

Brands improve AI visibility by publishing clear, structured content, maintaining consistent messaging across third-party platforms, and monitoring how AI systems describe them over time.

Does GEO replace SEO?

No. GEO builds on SEO fundamentals but requires additional strategies focused on semantic clarity, consistency, and AI-readable structure.

A guide to how the AI Act will impact marketing communications

Photo by Solen Feyissa on Unsplash