How Sembit Turned AI Search Into 28% of Revenue (Without Ads or Cold Email)
Interview: Getting Recommended by AI Instead of Chasing Leads
Behind the Agency Podcast with Shaun Davidson, VP at Sembit & Co-Founder of Zero Channel
Watch
Prefer the highlights? Key takeaways and summary below.
TL;DR – Key Takeaways
Sembit spent years stuck in referral-only growth and failed with ads and cold email.
One ChatGPT recommendation turned into a large project and changed their growth strategy.
Sampling AI models repeatedly (same prompt, 100+ times) revealed exactly who gets cited and why.
Earned media sources (Clutch, DesignRush, GoodFirms, even Yellow Pages) matter more than most people realize.
Picking a few high-intent prompts beats chasing hundreds of vague ones.
28% of Sembit’s last 9 months of revenue can be directly tied to LLM recommendations.
AI search is already pulling serious buyers away from Google.
Early adopters are getting outsized wins while most companies are still ignoring this.
Meet the Guest
Shaun Davidson is the VP of Sembit and the co-founder of Zero Channel, a platform focused on helping companies get recommended by AI tools like ChatGPT, Google AI Mode, and Gemini.
Sembit is a senior-level custom app development agency based in Portland, Oregon, working on complex web and mobile applications for enterprise and highly technical clients. After 17 years of growing mostly by referrals, Shaun turned an internal experiment into a new growth channel that now drives a meaningful share of revenue.
Episode Summary
1. From referrals to a hard ceiling
Sembit had all the classic “good agency” problems.
They were senior, technical, trusted, and had been around a long time. But growth depended almost entirely on word of mouth. Ads didn’t work. Cold email didn’t work. Even a $15k ad spend produced zero leads.
That forced Shaun to ask a different question: what happens if buyers stop searching the way we do?
2. The moment everything clicked
The breakthrough came when a prospect booked a call after spending 30 minutes talking to ChatGPT.
No shortlist. No comparison shopping. No deep website research.
“I had a half-an-hour conversation with ChatGPT and just went straight to your site.”
That deal turned into a large project and made something obvious: AI wasn’t just summarizing the web. It was making decisions for buyers.
3. Sampling the models instead of guessing
Rather than guessing how AI search worked, Shaun did something painfully simple and very unsexy.
He asked the same question 100 times.
Then he logged:
Who got mentioned
Which sources were cited
What patterns showed up consistently
This revealed a brutal truth: Sembit was invisible.
Once they saw that, the path forward wasn’t mysterious. The citations told them exactly what to fix.
4. The core method behind AI visibility
At a high level, the approach looks like this:
Pick a small set of high-intent, unbranded prompts.
Run them many times, not once.
Collect every citation the model uses.
Identify which sources show up often and actually move rankings.
Prioritize earned media and owned content that aligns with those sources.
Re-sample, re-test, and adjust continuously.
The citations themselves become the roadmap.
5. The common mistake everyone makes
The biggest mistake Shaun sees is prompt sprawl.
People generate hundreds of prompts and test each one once. That produces noise, not signal.
It’s better to dominate three well-chosen prompts than to barely show up in fifty.
If your competitors aren’t showing up in the results, the prompt is probably wrong.
6. Where founders and agencies still matter
This isn’t a “set it and forget it” system.
Human judgment still matters in:
Choosing prompts tied to real buying intent
Deciding which citations are worth pursuing
Implementing playbooks across earned and owned media
Expanding from local to regional, national, or international visibility
Agencies also have a natural advantage here: they can use the same system for themselves and productize it for clients.
7. What Zero Channel actually does
Zero Channel combines software and service.
The platform:
Scores visibility across AI search prompts
Tracks citations across models
Generates prioritized playbook tasks
Shows clear win conditions for each task
The service layer:
Helps select prompts tied to revenue
Produces actionable playbooks
Guides implementation
Reviews progress monthly and adjusts strategy
The goal isn’t more “visibility.” It’s confirmed leads and revenue tied to AI recommendations.
Notable Quotes
“We realized we were invisible.”
“The citations tell you what to do.”
“Running hundreds of prompts once is a very bad strategy.”
“Twenty-eight percent of our revenue came from language model recommendations.”
Learn More / Get in Touch
Visit → https://sembit.com
LinkedIn → Shaun Davidson
Message Shaun directly to book a conversation or demo
Want More Interviews Like This?
Subscribe to the show on YouTube
Subscribe to my weekly newsletter
