← All articles
AI Visibility 16 April 2026 14 min read

Case Study: How a Swiss B2B Company Reached Top 3 on Every AI Model

A real-world case study of a Swiss B2B data platform that achieved top-3 AI visibility across ChatGPT, Claude, and Perplexity — without backlinks, Wikipedia, or PR campaigns.

From Invisible to Recommended in Weeks

When we talk about AI visibility, most discussions focus on theory — what should work, what might work, what the best practices suggest. This case study is different. It documents what actually happened when a Swiss B2B company went from zero AI visibility to consistent top-3 recommendations across every major AI model.

The company is a Swiss B2B data platform — a SaaS product serving enterprises in the DACH region. We cannot name them directly, but we can share the details of what they did and the results they achieved, because their journey illustrates principles that apply to any Swiss B2B company.

This case study covers their starting point, the specific actions they took, the timeline of results across each AI platform, and the honest lessons — including the caveats that most GEO success stories leave out.

The Starting Point: Zero Visibility

When we first analysed this company's AI presence, the situation was typical of what we see with most Swiss B2B firms:

  • ChatGPT: Did not mention the company for any relevant query. Not in the top 5, not in the top 10 — simply absent.
  • Claude: Same result. Complete invisibility.
  • Perplexity: Occasionally surfaced the company website in search results but never cited it in the synthesised answer.
  • Google AI Overviews: The company appeared in traditional search results but was not included in AI-generated summaries.

This was despite the company having a functional website, a reasonable Google presence, and a product that was genuinely strong in its niche. The gap between their actual market position and their AI visibility was stark.

What They Did NOT Have

This is perhaps the most instructive part of the story. When this company achieved top-3 AI visibility, they did so without several things that GEO guides typically recommend:

  • No backlinks to speak of. The company had virtually no external backlink profile. No guest posts, no press coverage linking back to them, no directory backlinks beyond the basics.
  • No Wikipedia entry. The company did not have a Wikipedia page, nor a Wikidata entry.
  • No PR campaigns. They had not issued press releases, had no media coverage, and had not invested in any external communications strategy.
  • No social media presence. Their LinkedIn was minimal, and they had no meaningful community presence on forums or discussion platforms.

By every traditional GEO checklist, they should not have achieved what they did. And yet they did.

What They Actually Did: The Content-First Approach

The company's approach was radically simple: they focused entirely on making their website the best, most comprehensive answer to the questions their target audience would ask.

1. Deep, Factual Product Content

Every product page was rewritten to be specific, factual, and comprehensive. Not marketing copy — genuine technical documentation of what the product does, how it works, who it serves, and what makes it different. They included:

  • Specific capabilities with concrete examples
  • Integration details naming exact systems and protocols
  • Pricing transparency — not exact numbers, but clear tier descriptions
  • Technical specifications that engineers and procurement teams need
  • Use cases mapped to specific industries and company sizes

2. Comprehensive FAQ Content

They built an extensive FAQ section that addressed every question a buyer might ask — not generic questions, but the specific, context-rich queries that Swiss B2B buyers actually type into AI tools. Each answer was detailed, factual, and stood on its own as a useful piece of content.

3. Technical Foundations

The website was technically clean in ways that matter for how AI models crawl and process content:

  • Server-side rendered — no JavaScript dependency for core content
  • Clean heading hierarchy (H1 to H3) on every page
  • Fast page load times (under 2 seconds)
  • Complete schema markup (Organization, Product, FAQ)
  • An llms.txt file summarising the company and its offerings
  • All AI crawlers explicitly allowed in robots.txt

4. Structured Data and Machine-Readable Identity

They implemented comprehensive llms.txt and schema markup that gave AI models a clean, authoritative source of truth about the company. Every fact was consistent — the same company description, the same product names, the same value proposition — across every page and every structured data element.

The Results

Within weeks of implementing these changes, the results were measurable:

  • Perplexity: The company appeared in the top 3 for every relevant query tested. Perplexity cited their website directly, often as the primary source.
  • ChatGPT: When triggered to search (which happened for most queries including "2025" or "current"), the company appeared consistently in the top 3. For training-data-based answers, improvement came more gradually as new model updates incorporated their content.
  • Claude: Similar to ChatGPT — strong results when web search was triggered, with training-data results improving over subsequent weeks.
  • Google AI Overviews: The company was regularly featured in AI-generated summaries for their key terms.

The improvement was not marginal. They went from zero mentions to consistent top-3 placement across all major AI platforms. For their primary category query, they achieved the number-one recommendation on three of the four platforms.

Why It Worked: The Product Quality Factor

Here is the honest part that many GEO guides gloss over: this worked because the product was genuinely excellent. The company was not trying to game AI models into recommending something mediocre. They were making it possible for AI models to discover and accurately describe something that was already the best answer.

AI models are remarkably good at distinguishing substance from marketing. When the website content was specific, factual, and comprehensive, the models could evaluate it on merit — and the merit was strong. The content did not need to persuade; it needed to inform. And when the information was genuinely impressive, the models recommended accordingly.

This is a critical lesson: GEO cannot make a weak product look strong. But it can — and this case proves it — make a strong product visible.

Key Takeaways for Swiss B2B Companies

1. Website Content Quality Is the Foundation

No amount of backlinks, PR, or social media can substitute for excellent website content. If your product pages read like marketing brochures instead of comprehensive product documentation, AI models have nothing substantive to work with. Start with your website content before investing in anything else.

2. Technical Foundations Are Non-Negotiable

Server-side rendering, clean HTML, proper schema markup, fast page speeds, and open access for AI crawlers — these are not optional extras. They determine whether AI models can even read your content. A brilliant product page that requires JavaScript to render is invisible to most AI crawlers.

3. External PR Is a Bonus, Not a Requirement

This case study proves that a company can achieve top-tier AI visibility without backlinks, Wikipedia, or press campaigns. These external signals help — press releases accelerate visibility and citations compound over time — but they are not prerequisites. If your budget is limited, invest in website content first.

4. Consistency Across All Signals Matters

The company's information was identical everywhere — website, llms.txt, schema markup, directory listings. This consistency gave AI models high confidence. Any discrepancy — a different company description on your LinkedIn versus your website, a product name that varies across pages — reduces that confidence.

5. Measurable Results Come Faster Than Expected

The conventional wisdom says AI visibility takes months. This company saw results within weeks. Perplexity, which always searches the web in real time, responded almost immediately. ChatGPT and Claude followed as their search features encountered the updated content. The 30-day timeline in our GEO roadmap is realistic, not optimistic.

The Detailed Timeline: Week by Week

Here is a more granular view of how the company's AI visibility evolved over time:

Week 0 (Baseline)

Zero mentions across all platforms. Ten buyer-relevant prompts tested across ChatGPT, Claude, Perplexity, and Google AI. The company appeared in exactly zero of the forty total responses. Three competitors dominated the responses, each appearing in 60-80% of queries.

Week 1 (Technical Setup)

The company deployed their llms.txt file, implemented schema markup, updated robots.txt to allow all AI crawlers, and registered with Bing Webmaster Tools. No content changes yet. No improvement in AI visibility — this was expected, as the technical changes needed time to be processed by crawlers.

Week 2 (Content Overhaul)

Product pages were rewritten with specific, factual content. FAQ sections were added. The About page was transformed from a marketing narrative into a fact-rich company profile. The company published their first two expert articles. By end of week 2, Perplexity began citing the company website in two of ten test prompts — the first sign of progress.

Week 3 (Early Traction)

Perplexity citations increased to five of ten test prompts. The company appeared in Google AI Overviews for their primary category keyword. ChatGPT mentioned them once when web search was triggered by a prompt containing "2025." Claude remained silent.

Week 4 (Inflection Point)

Perplexity consistently cited the company in seven of ten prompts, often as a top-three recommendation. ChatGPT mentioned them in four of ten prompts (all search-triggered). Google AI Overviews included them for three key terms. Claude mentioned them for the first time in two prompts where web search was triggered.

Week 6-8 (Consolidation)

By week eight, the company appeared in the top three recommendations on Perplexity for every relevant prompt tested. ChatGPT included them in six of ten prompts. Google AI Overviews featured them consistently. Claude mentioned them in three of ten prompts. The trajectory was clear and accelerating.

Week 12 (Sustained Position)

Three months after starting, the company held consistent top-three positions across all platforms for their primary category queries. More importantly, the results were stable — re-running the same prompts week after week produced consistently positive results, indicating that the AI models had developed lasting knowledge of the company.

What They Would Do Differently

In retrospect, the company identified several things they would change if starting over:

  • Register with Bing on day one. They waited until week one to register with Bing Webmaster Tools. Since Bing indexation takes one to two weeks, this delayed their ChatGPT visibility by two weeks. Starting with Bing on day one would have accelerated their timeline.
  • Publish expert content earlier. Their expert articles were not published until week two. If they had prepared drafts before starting the technical setup, they could have published simultaneously with their technical changes — giving AI crawlers fresh, optimised content to index immediately.
  • Issue a press release. Despite achieving top-three visibility without PR, the company believes a press release in week two would have accelerated their results, particularly on ChatGPT and Claude where third-party authority signals carry extra weight.
  • Test more prompt variations. Their initial prompt set of ten was sufficient for baseline measurement but too narrow for comprehensive gap analysis. They later expanded to twenty-five prompts across both English and German, which revealed additional content gaps they had not initially identified.

Applying This Case Study to Your Company

To apply these lessons to your own Swiss B2B company, use this self-assessment framework:

Product Quality Assessment

Honestly evaluate your product or service against competitors. If you are genuinely the best option in your niche — or one of the top two to three — GEO can make your quality visible to AI models. If you are a mid-tier option, GEO will ensure you are accurately represented, but you should not expect to outrank objectively superior competitors. AI models are remarkably good at assessing quality from content signals.

Content Readiness Assessment

Evaluate your website content against these criteria:

  • Does your product page describe specific capabilities, integrations, and technical specifications — or is it marketing copy?
  • Do you have FAQ content that addresses the specific questions buyers ask?
  • Are your key facts (founding year, employee count, client count, pricing) clearly stated on your website?
  • Is your content available in both English and German?

If most answers are "no," your first step is content overhaul — which is exactly where the case study company started.

Technical Readiness Assessment

  • Does your website render key content without JavaScript? (Test by disabling JS in your browser.)
  • Do you have schema markup (Organization, Product, FAQ) implemented?
  • Is your robots.txt configured to allow GPTBot, ClaudeBot, and PerplexityBot?
  • Are you registered with Bing Webmaster Tools?
  • Do you have an llms.txt file deployed at your domain root?

If most answers are "no," start with technical foundations before investing in content.

The Honest Caveat

We share this case study because it demonstrates what is possible. But we want to be transparent: these results are not guaranteed for every company. This company had a genuinely strong product in a clearly defined niche. The content accurately reflected real capabilities. AI models recommended them because the recommendation was correct.

If your product is average in a crowded field, no amount of GEO will make AI models rank you first. What GEO will do is ensure AI models can find you, understand you, and accurately represent your offering. The final recommendation still depends on the quality of what you actually deliver.

That said, most Swiss B2B companies are better than their AI visibility suggests. The gap is not in their product — it is in how well that product is communicated to AI models. Closing that gap is where GEO delivers transformative results.

Frequently Asked Questions

Can these results be replicated in a more competitive niche?

Yes, but the timeline may be longer and the effort greater. This company operated in a niche with moderate competition — three to five serious competitors. In highly competitive categories (e.g., general IT consulting, generic cloud services), achieving top-three placement requires more effort because there are more companies with strong web presences competing for the same AI recommendations. However, the principles remain the same: specific, factual content; technical accessibility; and consistent information across sources. In competitive niches, the differentiating factor is often the depth and specificity of content — not volume, but quality. A company that publishes ten detailed, scenario-specific guides will outperform one with fifty thin pages.

Did the company use per4mx during this process?

Yes. per4mx was used for baseline measurement, weekly visibility tracking, gap analysis, and competitive monitoring throughout the process. The tool provided the data that guided content decisions and measured the impact of each change. While it is possible to achieve similar results with manual monitoring, per4mx significantly reduced the time spent on measurement and provided structured insights that accelerated decision-making.

How much did this entire effort cost the company?

The direct costs were minimal: per4mx subscription (CHF 79/month), no press release costs, no advertising spend, no agency fees. The primary investment was time: approximately 40-50 hours over the first month, concentrated in content writing and technical implementation. This was handled by a team of two — one marketing lead and one developer. After the initial sprint, ongoing maintenance required approximately 8-10 hours per month for content updates and monitoring review. Total first-year cost: under CHF 5,000 including tool subscription and team time. Given that the company attributed multiple six-figure enterprise deals to their improved AI visibility, the ROI was substantial.

Is it possible to achieve these results without any technical skills?

The content-related work (rewriting pages, creating FAQs, writing expert articles) requires no technical skills — only subject matter expertise and clear writing. The technical work (schema markup, llms.txt deployment, robots.txt configuration) requires basic web development skills. If you do not have a developer on your team, a freelance web developer can handle all technical tasks in four to six hours. Many CMS platforms (WordPress, Webflow) also offer plugins or built-in features that simplify schema markup and static file deployment.

Ready to take action?

Check your AI visibility for free

See how ChatGPT, Claude, Perplexity, and Gemini describe your company today. Get a free visibility report in minutes.