How to Get Leads From ChatGPT: A Real 90-Day Case Study

Most local businesses are already getting customers from ChatGPT, Perplexity, and Gemini. They just cannot see it. We tracked it for one of our clients, RWA Center, a Beverly Hills wellness clinic, for 90 days. Here is what the data showed, and the four moves any local business can copy.

People used to Google “IV therapy near me” and scan ten links. Now a growing share ask an AI assistant a full question and get one answer that sometimes names a business and links to it. The problem: that click often lands in your analytics as “direct” traffic, not as ChatGPT. So the channel is real, the visitors are real, and for most owners the reporting is blind.

It is not fringe. Across most business websites in 2026, AI assistants are roughly 1% of traffic, yet only about 14% of marketers actually track it. That gap is the opportunity.

What 90 days of data showed

We pulled 90 days of analytics for RWA Center (rwacenter.com), a mid-sized regenerative medicine and aesthetics clinic in Beverly Hills, and isolated every visit from a known AI assistant.

These visitors did not bounce. They produced 63 booking and call clicks, 21 started consultation forms, and 6 confirmed enquiries. A channel under 1% of traffic delivered real, high-intent leads for services that run from a few hundred to over a thousand dollars each.

Is 260 sessions small? Yes, and we will not pretend otherwise. But independent 2026 studies repeatedly find AI-referred visitors convert several times higher than ordinary search, because they arrive pre-qualified by the assistant’s answer. AI is the fastest-growing, highest-intent channel most competitors are not even watching.

The funnel it feeds

To see why it matters, look at the funnel. RWA Center’s website forms produced roughly 570 leads, around 45 booked appointments, with a typical visit of a few hundred dollars and an average above $2,000 once premium packages are included, adding up to tens of thousands in booked revenue.

Here is the connection. The pages AI assistants send people to, and the pages that generate booked leads, are largely the same pages: the in-depth treatment and peptide guides and the clear service pages. We are not claiming AI produced that revenue directly, it came from every channel combined. What the data shows is cleaner: AI is already steering people to the exact content that the clinic has proven turns visitors into paying patients.

One strategy, two payoffs

RWA Center never ran a separate “get cited by AI” campaign. It committed to one thing, publishing genuinely helpful, trustworthy content. That paid off twice.

Payoff one: AI assistants started citing it.

Over six months, the content earned citations across AI Overviews, Perplexity, Gemini, Copilot, Grok, and ChatGPT, with most of the growth landing after the content strategy shifted into gear about four months ago.

Payoff two: Google visibility climbed too.

The same content Google rewards. Over the same window, organic clicks and impressions trended steadily up.

We are not claiming the citations caused the Google growth, or the reverse. Both are downstream of the same root cause, helpful content. One investment, two compounding returns.

What actually went into it

The results above did not come from one trick. They came from a steady, several-month program across a few fronts. Kept high-level, here is the shape of it:

• Content audit and cleanup. Fixed pages competing for the same searches, thin pages, and gaps where demand had no matching page, before publishing anything new.
• New, genuinely useful content. In-depth guides written to inform, not sell, mapped to the real questions people ask. These now earn both AI citations and Google rankings.
• Metadata and on-page optimization. Titles, descriptions, and page structure rewritten to be clear and aligned to how people actually search.
• Discoverability improvements. Technical fixes so search engines and AI crawlers can find and trust the site: cleaner URLs, internal linking, structured signals.
• Bot access control. Deliberate choices about which crawlers reach what, so the assistants you want citing you can read your content. Getting this wrong is a common, invisible reason businesses never appear in AI answers.
• Credibility and tracking groundwork. Consistent business info, clearer trust signals, and analytics set up to measure the AI traffic most setups miss entirely.

One quick way to see this work pay off is an “agent-readiness” check, a free scan (the one below is from Cloudflare) that grades how easily AI agents and crawlers can discover, read, and trust a site across exactly these areas: discoverability, content readability, and bot access. After the work, the clinic passes the checks that matter.

Four moves you can make this month

1. Make the traffic visible. Standard analytics will not separate AI traffic out of the box. Group visits from ChatGPT, Perplexity, Claude, Gemini, and Copilot into their own channel. You cannot manage what you cannot see.

2. Audit how the assistants describe you. Ask ChatGPT, Perplexity, and Gemini the questions a real customer would ask about your category and city. Do you get mentioned? Is it accurate? A 20-minute reality check.

3. Turn your best service into a real guide. Pick your highest-value service and write one honest page: what it is, who it suits, what to expect, the real limitations, and why your team is qualified. Write to inform, not to sell. This is the content the data shows AI prefers to cite.

4. Make your credibility machine-readable. Consistent business info everywhere, real credentials, genuine reviews, clear authorship. It helps AI trust you and reassures the humans who land afterward.

Those four steps get you visible and give you data. Going from “occasionally mentioned” to “consistently recommended across every assistant” is the harder, compounding work, and it is what we do for clients.