From $0 to 7-Figure Pipeline from ChatGPT in 10 Mos
Case study · AI visibility
From $0 to 7-figure pipeline from ChatGPT in 10 months.
The challenge: rebuilding search visibility after AI disrupted everythingBetween 2023 and 2025, generative AI quietly rewired how people research, compare, and choose vendors. Traffic splintered across tools like ChatGPT, Perplexity, Claude, and Gemini, and the old “Google-only” playbook stopped telling the full story. My role was to keep the brand discoverable, credible, and easy for AI models to recommend in those conversations.
Why this matters
AI visibility requires a different kind of storytelling.
In the AI era, visibility isn’t just about ranking for keywords anymore. Machines have to understand the story of who you are: your capabilities, your differentiators, and the different ways those strengths show up across real campaigns.
When that story is clear and consistent, AI tools do something powerful: they start sending you the right people. Not just traffic. Qualified, ready-to-convert prospects who match your ideal customer profile.
LLMs help buyers narrow hundreds of agency options down to a handful of strong fits.
If a model understands your true strengths, you get surfaced at the exact moment someone needs the thing you’re genuinely best at.
Context + approach
When search stopped being just Google.
What the data showed
- I began tracking LLM-driven traffic in 2024. By year-end, it made up 0.67% of sessions.
- In 2025, traffic from ChatGPT, Perplexity, Claude, and Gemini rose to ~4% of all site traffic, peaking near 5%.
- LLMs provide no referrers or query data, so traditional analytics couldn’t explain how people were finding us.
- Path exploration revealed distinct behaviors: Perplexity acted like a research engine, while ChatGPT acted like a decision engine.
- 20% of Perplexity traffic landed directly on case studies; ChatGPT users arrived with near-BOFU intent.
“ChatGPT compresses the funnel. People walk in wondering and walk out ready to buy. I’ve never seen anything move decision-making that fast.”
My internal observation from LLM path explorationWhat needed to change
- Treat AI tools as distinct discovery channels with unique intent behaviors, not extensions of SEO.
- Move from keyword-first SEO to entity-first content, so LLMs could clearly interpret who we are and when to recommend us.
- Rewrite and restructure pages to be machine-legible, with explicit definitions, roles, and relationships.
- Add AI-aware metadata and schema to strengthen model confidence in associating us with specific problems and services.
- Run continuous path exploration to see how LLM-driven visitors moved across the site and where journeys diverged by platform.
Same audience, different engines
Two platforms. Two completely different buyers.
Path exploration showed that each LLM sent people with a different job to do, which meant each one needed different content waiting for them.
The research engine
Research-heavy visitors. 20% of Perplexity traffic landed directly on case studies, looking for proof before going any further.
The decision engine
Visitors arrived with near-BOFU intent, behaving like prospects already close to a decision. ChatGPT compresses the funnel.
How it played out
Map the traffic. Then teach AI who we are.
Figure out how people actually find us now
The first step was reconstructing behavior without traditional analytics. With no referrer or prompt data from LLMs, I traced user journeys manually to understand which pages they hit first and what patterns emerged.
- Segmented traffic from ChatGPT, Perplexity, Claude, and Gemini inside analytics dashboards.
- Ran path exploration to understand LLM-driven entry points and navigation patterns.
- Confirmed that Perplexity visitors were research-heavy, often landing on case studies.
- Confirmed that ChatGPT visitors arrived closer to a decision, behaving like BOFU prospects.
- Identified question clusters and themes that surfaced repeatedly across platforms.
Teach AI who we are and when to recommend us
Once I understood how each LLM behaved (and this is client-specific, based on the groundwork we’d laid over the years with content), the focus shifted to restructuring the site so AI models could accurately interpret our expertise, services, differentiators, and ideal-fit use cases.
- Rebuilt all case studies using structured, modular storytelling designed to clarify outcomes for humans and machines.
- Created AI-first content hubs with separate pages for each tactic, capability, and differentiator.
- Developed machine-training pages that isolated single concepts for precise model interpretation.
- Embedded hidden microdata fields to give models context without overwhelming human readers.
- Rewrote key landing pages in entity-first, machine-readable language and added structured data to reinforce meaning.
Key learning
AI changed what “being discoverable” means.
The biggest insight from rebuilding visibility for AI-driven discovery.
Optimizing for AI improved all machine visibility.
One unexpected insight: optimizing content for LLMs didn’t just improve recommendations inside ChatGPT or Perplexity. It strengthened our visibility across all machine-driven discovery systems. By restructuring pages to be more readable, explicit, and machine-legible, we saw a dramatic increase in traffic from non-Google search engines.
Optimizing for LLMs makes you easier for any machine to understand, not just AI assistants. LLM-first content architecture created a rising-tide effect across Bing, niche engines, research tools, and traditional search. The shift from keyword optimization to machine comprehension translated directly into measurable revenue.
Do AI systems know when to recommend you?
Find out what ChatGPT, Perplexity, and the rest actually say about your company, and what it would take to get surfaced at the right moment.