How Generative AI Is Changing Online Behaviour
Learn how AI is reshaping search behaviour, and the framework to get your brand seen and cited.
Are you wondering why your impressions are climbing but your traffic is heading in the opposite direction?
You’re not imagining it. The way people search, evaluate, and ultimately choose which brands to trust is undergoing its most significant shift in a quarter of a century. Users are increasingly bypassing traditional search results entirely — turning to AI assistants for answers, comparisons, and purchasing decisions. And for marketing leaders responsible for organic visibility, the implications are enormous.
In a recent live session, Charlie Marchant (CEO of Exposure Ninja) sat down with Fernando Angulo (Senior Market Research Manager at Semrush*) to unpack exactly what’s changing — backed by billions of tracked keywords, enterprise case studies, and a practical framework your team can act on immediately. Fernando brings rare insight from Semrush’s enterprise MCP pilot, including findings from clients such as eBay, Zalando, and Yext.
This article distils the key findings from that session into a practical guide for marketing leaders navigating the shift to AI Search Optimisation.
Executive Summary
The transition from keyword-based search to conversational AI is no longer theoretical — it’s measurable, accelerating, and already reshaping how brands are discovered. Here’s what the data shows:
- AI Overviews now appear on more than 25% of all Google search queries, and that number is growing every month. The traditional “position one” ranking is being pushed further down the page — often below AI-generated answers, YouTube videos, ads, and Reddit discussions.
- ChatGPT referral traffic has grown by over 200%, even as the platform’s own user growth has begun to plateau. Users aren’t just asking questions — they’re clicking through to websites from AI-generated answers at an increasing rate.
- 65–85% of all ChatGPT prompts now use conversational language rather than traditional keyword-based queries. Users are asking AI to explain, compare, and recommend — not typing three-word search terms.
- Only 20% of AI platform users fully trust the results they receive, creating what Fernando describes as a “trust paradox” — people use AI tools operationally but still verify strategically.
- Semrush increased its own AI visibility from 10% to 35% in just two months by implementing the framework shared in this session.
The businesses that act on these shifts now will establish the kind of early-mover advantage that becomes very difficult to replicate once competitors catch up.
What Is Actually Happening in AI Search Right Now?
For roughly 25 years, search worked the same way. You had a need, you translated it into keywords, you typed those keywords into a search box, and you chose from a list of results. The first three positions captured most of the clicks. The model was simple, predictable, and — for businesses that understood SEO — highly profitable.
That model is breaking.
As Fernando put it during the session: “25 years. Gone. Nobody’s saying ‘Google it’ anymore when they’re going to do their research.”
The shift has moved through three distinct phases in rapid succession. Google’s Search Generative Experience gave way to AI Overviews, which have now evolved into full AI Mode. Three names, one direction — AI is becoming the primary interface between users and information.
What’s particularly striking is the scale. Semrush’s data, drawn from billions of tracked keywords across more than 190 countries, shows that 40% of US desktop users are now visiting AI tools regularly, with around 20% using them ten or more times per month. And projected adoption figures from Statista suggest that 68% of US adults will be using generative AI tools by the end of 2030.
This isn’t a niche trend for early adopters. It’s mass adoption — and it’s happening faster than most marketing teams have adjusted for.
How Has User Behaviour Shifted From Keywords to Conversations?
One of the most revealing data points Fernando shared relates to how people are actually using these platforms. The shift isn’t just about where people search — it’s about how they search.
In mid-2025, the average ChatGPT prompt was around 28–30 words. Think of something like: “Rewrite this document with a corporate tone for my CEO presentation next Tuesday.” By February 2026, that figure had dropped to roughly 10 words. Users have learned that they don’t need elaborate prompts — they describe what they want more directly.
More importantly, the type of language has shifted dramatically. Traditional keyword-style queries — things like “best coffee NYC” or “iPhone 15 price” — still account for 15–35% of ChatGPT prompts. But conversational, intent-first queries now make up 65–85% of all prompts. These are questions like:
- “Explain this to me like I’m a designer”
- “What should I do about this situation?”
- “Help me think about this decision”
This matters enormously for content strategy. If your content is built around matching short-tail keywords, it’s optimised for a search behaviour that represents a shrinking minority of how people actually look for information. The businesses winning in this environment are the ones creating content that answers the way people now ask — conversationally, contextually, and with intent at the centre. If you’re rethinking your approach to content, our guide to AI Search Optimisation strategy covers exactly how to restructure for this shift.
There’s also a fundamental change in what Fernando calls “cognitive shift” — the move from critical thinking to algorithmic trust. In the old model, users compared multiple results, weighed sources, and made their own judgement. In the new model, they receive a single synthesised answer and — increasingly — simply accept it. Brand visibility is no longer measured by where you rank. It’s measured by whether AI mentions you at all.
Why Are Zero-Click Results Accelerating?
Zero-click search isn’t new. But the scale at which it’s now happening — and the reasons behind it — have changed fundamentally.
When you ask Google a question today, the experience looks nothing like it did even 18 months ago. You’ll typically see an AI Overview consuming the top portion of the page, followed by YouTube videos, ads, Reddit threads, and news results. The traditional organic listing that used to sit in “position one”? It’s now often below the fold entirely.
The result is that users are getting complete answers without clicking on anything. AI Overviews appear on more than a quarter of all search queries, and that proportion is growing. For informational queries — which remain the dominant intent type triggering AI results — the answer is delivered directly in the search interface.
Fernando was direct about the implications: “Nobody is clicking anymore to find information, to gather information about a product or service on Google — because there are no clicks.”
For the industries receiving the most AI-driven traffic, the pattern is clear. Online services leads the pack at around 10 million ChatGPT users per month, followed by mass media, publishing, computer software, and education. These are all knowledge-acquisition industries — exactly the sectors where conversational AI is replacing traditional search most rapidly.
This doesn’t mean traditional SEO is irrelevant. It means the game has expanded. As Fernando framed it: “AI is not replacing discovery. It’s expanding it.” The blue links are still there — they’ve just got a different role now. And if your brand isn’t appearing in the AI-generated layer above them, you’re invisible to a growing proportion of your potential customers.
What New Metrics Should Marketers Be Tracking?
One of the biggest practical challenges marketing leaders face is measurement. Traditional SEO gave us a well-understood set of KPIs — search volume, click-through rate, conversion rate, bounce rate. Those metrics still matter, but they don’t capture what’s happening in AI search.
Semrush has developed three new KPIs that their own teams — and their enterprise clients — are now tracking:
Brand visibility measures how often your brand name appears in AI-generated content. This can be either a mention (your brand name appears in the answer) or a citation (the AI links to your content as a source). Both count. Both matter.
Share of voice tracks how frequently your brand is mentioned within a specific topic area compared to your competitors. If someone asks ChatGPT or Google’s AI Mode about the best tools in your category and your competitors appear but you don’t, your share of voice is zero — regardless of how well you rank organically.
Sentiment evaluates whether your brand mentions are positive, neutral, or negative relative to competitors. It’s not enough to be mentioned — the context in which AI presents your brand shapes user perception just as powerfully as the mention itself.
Charlie added an important practical note during the session: these metrics are surfacing in new tools and dashboards across the industry. Semrush One* — their unified platform — now includes an AI visibility checker that shows how different LLM platforms perceive your website, and you can compare that against your traditional SEO performance to spot gaps.
If you want to learn how to build a reporting framework around these new metrics, our step-by-step guide on how to track and improve your brand’s LLM visibility walks through the entire process.
How Can Brands Build AI Visibility?
This is where Fernando’s presentation shifted from diagnosis to prescription — and the framework he shared is one of the clearest we’ve seen.
It boils down to two objectives: get seen and be trusted.
Getting Seen: The Sentiment Game
AI platforms don’t just crawl your website. They pull information from six distinct source types — and each one maps to a different team inside your organisation:
- Review platforms for product comparisons — that’s your customer success team
- Reddit threads and forums for pricing discussions and complaints — that’s your social and community team
- Developer forums for implementation details — that’s support and community building
- News sites for company credibility — that’s public relations
- Support documentation for feature explanations — that’s customer education
- Your pricing page for commercial information — that’s product marketing
This is why Fernando was emphatic that AI visibility is a company-wide effort, not a task for the SEO team alone. Every one of those source types feeds into how AI platforms understand, recommend, and cite your brand. If your reviews are negative, your Reddit presence is non-existent, and your pricing isn’t published online, the AI has gaps in its understanding of you — and it will either skip you or fill those gaps with information you can’t control.
The practical actions for getting seen include building presence on the right review sites, participating actively in community discussions on platforms like Reddit and Quora, engineering user-generated content and social proof, and activating customer and employee advocacy programmes. As Fernando explained: “Most users who have a bad experience go to social media. Nobody is pushing users to share their good experience. You need to do that.”
If you’re looking for a structured process to identify exactly where your brand is and isn’t showing up, our guide to running an AI Search Optimisation audit gives you a step-by-step framework.
Being Trusted: The Authority Game
Getting mentioned is only half the equation. To earn citations — where AI links back to your content as a source — you need to establish trust signals that LLMs recognise and reward.
Fernando outlined several specific trust-building strategies:
Optimise your site for AI platforms directly. This means ensuring your schema markup is comprehensive and accurate. Fernando made the point that LLMs are, at their core, machines — and structured data is their native language. Charlie reinforced this from her experience reviewing thousands of websites: “I don’t think I’ve ever seen one with a perfect schema set up across the site — there are about a hundred different types, and setting it up correctly takes time.”
Get your Wikipedia and Knowledge Graph entries in order. Fernando described Wikipedia as being “relevant again after three decades” — because LLMs treat it as a foundational source for understanding entities. If your company has a Wikipedia page, it needs to be accurate, current, and comprehensive. If it doesn’t, the Knowledge Graph becomes even more important.
Publish transparent pricing. This was one of the most counterintuitive recommendations — and one of the most powerful. Many businesses, particularly in B2B, deliberately hide their pricing behind a “contact us” form. But LLMs need factual data to generate accurate answers. If your pricing isn’t publicly available, AI platforms either skip you or hallucinate a figure — neither of which is good for your business.
Keep your Google Business Profile current. When asked directly whether Google Business Profiles were becoming less important, Fernando was unequivocal: “By far, no. Your Google Business Profile is important more than ever. That’s a source of information for LLM platforms.” Charlie confirmed that Business Profiles appear constantly in Google’s AI Mode results, particularly for local queries.
For a deeper look at how to structure content specifically for AI consumption, our topic research guide for AI platforms covers the formats and structures that LLMs favour most.
How Is Semrush Measuring AI Search Success Internally?
What makes Fernando’s framework particularly credible is that Semrush has applied it to their own marketing — and shared the results publicly.
The company implemented a coordinated, company-wide AI visibility strategy starting in June of last year. The changes spanned content, community, review platforms, partnerships, and structured data. Within two months, their AI visibility score rose from 10% to approximately 35% — and it’s continued to climb since.
Internally, Semrush uses a five-stage process:
- Research — Listen to your customers and understand the language they use across all touchpoints.
- Architecture — Design your site navigation to mirror customer mental models and search behaviour.
- Content — Align all internal teams’ content output so it creates a consistent signal for AI platforms.
- Distribution — Ensure your brand appears across all six source types that LLMs draw from.
- Measurement — Track brand visibility, share of voice, and sentiment as your primary AI KPIs.
Fernando also made an important observation that should reassure any marketer feeling overwhelmed by this shift: “That sounds, that feels, that smells like all good SEO — because it is. These are good practices. But right now they are relevant more than ever, because LLMs are not going to take three to six months to recognise you. They recognise you immediately.”
The investment in solid SEO fundamentals isn’t wasted — it’s the foundation. AI Search Optimisation builds on those fundamentals and extends them into the channels and formats that AI platforms actually prioritise.
What Does Winning AI Visibility Look Like in Practice?
To illustrate what best-in-class AI visibility looks like at enterprise scale, Fernando walked through a detailed case study of Salesforce.
Salesforce has built dedicated content pages for every industry vertical it serves — including automotive, financial services, and healthcare — specifically designed to be discovered by AI agents. These aren’t generic landing pages. They’re contextually rich, partner-specific pages that create explicit associations between Salesforce and its clients. For example, Salesforce has a unique page for Mercedes-Benz that establishes the relationship, the customer experience, and the CRM context — all signals that LLMs can parse and reference.
The company has also published pricing for every product and vertical — something Fernando noted they had deliberately avoided for over a decade. The reason for the change? LLMs need factual pricing data to answer user queries accurately. Without it, the AI either skips Salesforce in its recommendations or risks generating incorrect figures.
Salesforce reinforces all of this with YouTube content produced alongside partners and affiliates, creating multiple content types across multiple channels — exactly the distributed presence that AI platforms reward.
The results speak for themselves. Using Semrush’s AI visibility checker, Salesforce shows strong performance across ChatGPT, AI Overviews, and AI Mode — the three environments where high-value buyers are increasingly making decisions.
Fernando closed with three principles that apply regardless of company size:
- You can’t have AI visibility without organisational clarity. This has to be a cross-functional effort — marketing alone can’t own every source type that AI platforms draw from.
- You can’t have organisational clarity without audience obsession. Understanding where your audience searches, how they ask questions, and what they trust is the starting point for everything else.
- Your site architecture proves whether you have it or not. Structure isn’t a technical afterthought — it’s how AI platforms determine whether your content is worth citing.
If you’d like to see how other enterprise businesses are approaching AI strategy, or how Fortune 500 companies are dominating AI search, those resources go deeper into the patterns the top performers are following.
Next Steps
The shift from traditional search to AI-driven discovery is measurable, accelerating, and already reshaping how buyers find and evaluate brands. Here’s how to start responding — broken down by timeframe so your team can act progressively rather than trying to do everything at once.
This Week
- Run an AI visibility check. Use Semrush One’s* free AI visibility checker to see how ChatGPT, Perplexity, Claude, and Google’s AI Mode currently perceive your brand. Compare the results against your organic SEO performance to identify gaps.
- Ask AI about your brand. Open ChatGPT, Gemini, and Perplexity and ask the questions your customers are asking — “What’s the best [your category] for [your audience]?” Note whether your brand appears, whether it’s cited or just mentioned, and how it’s described relative to competitors.
- Audit your pricing transparency. Check whether your pricing information is publicly accessible. If it’s hidden behind a contact form, that’s a gap AI platforms can’t fill accurately — and they’ll either skip you or guess.
Next 30 Days
- Map your six source types. Review platforms, Reddit and community discussions, developer forums, news coverage, support documentation, and your pricing page. Identify which ones are strong, which are weak, and which are absent entirely.
- Review your schema markup. Check every schema type currently on your site for accuracy and completeness. There are roughly a hundred different types — most businesses are using fewer than ten, and some of those may be outdated or incorrect.
- Update your Google Business Profile and Wikipedia presence. Ensure all details are accurate and current. These are foundational entity signals that LLMs draw on heavily.
Next 90 Days
- Build a cross-functional AI visibility programme. Bring together customer success, PR, product marketing, community management, and content teams around a shared set of AI KPIs — brand visibility, share of voice, and sentiment.
- Launch customer and employee advocacy initiatives. Create structured programmes that encourage positive reviews, community participation, and genuine UGC across the platforms AI draws from.
- Develop an AI search strategy that aligns your content, technical infrastructure, and distribution channels specifically for AI visibility — not just traditional organic rankings.
If you’d like a professional assessment of where your brand stands and what to prioritise first, request a free review of your website and marketing from the Exposure Ninja team.
In Conclusion
The data is clear: the way people search, evaluate, and choose brands has fundamentally changed. AI Overviews appear on more than a quarter of all search queries. Conversational prompts have overtaken keyword-based searches. And referral traffic from AI platforms is growing at a pace that most marketing teams haven’t yet accounted for.
But this isn’t a story about the death of SEO. It’s a story about its expansion.
The fundamentals — great content, strong authority, genuine expertise — matter more than ever. What’s changed is where and how those signals need to appear. Your website alone is no longer enough. AI platforms are pulling from review sites, community discussions, structured data, news coverage, and pricing pages to build the answers they serve to users.
The marketing leaders who respond now — by auditing their AI visibility, aligning their teams around new metrics, and building the distributed presence that AI platforms reward — will establish an advantage that compounds over time. The ones who wait will find that their competitors have already taken the space.
As Fernando put it: “LLMs are not going to take three to six months to recognise you. They recognise you immediately. That’s the power of AI SEO.”
The shift is here. The data confirms it. And the opportunity is open — but it won’t stay that way for long. You can watch the full webinar on demand to get every data point, case study, and audience question covered in the live session. Or, if you’re ready to take action, explore the best AI Search Optimisation strategies for 2026 and start building your plan today.
*Some links within this article are affiliate links for which Exposure Ninja receives a fee for promoting (these links are not sponsored).
