Episode 63: 28% of Revenue Came from AI Recommendations

Feb 6

Episode 63 | Behind the Agency Podcast with Shaun Davidson, VP at Sembit and Co-Founder of Zero Channel

“A prospect spent 30 minutes with ChatGPT, got our recommendation, searched us on Google, and booked. No comparison shopping.”

Watch

TL;DR -- Key Takeaways

28% of Sembit's revenue over 9 months came from prospects who found them through LLM recommendations. This is not a future trend. It is happening now, and most agencies have no idea whether they are visible.
Citation-first methodology: test your target prompts 100+ times to understand how AI models reference (or ignore) your agency. Without a baseline, you are optimizing blind.
Directory profiles carry outsized weight. Clutch, DesignRush, and GoodFirms profiles are heavily represented in AI model training data. Complete profiles with reviews, case studies, and accurate service descriptions improve your chances of appearing in AI-generated recommendations.
High-intent focused prompts outperform scattered queries. Optimize for the specific questions your ideal client would ask an AI, not generic category terms.
Actionable test you can run today: Ask ChatGPT the prompt your ideal client would use. Run it 5 times. If your agency never appears, you have found the gap.
The buying journey is changing. Prospects are using AI to shortlist agencies before they ever open Google. If you are invisible to LLMs, you are invisible to a growing share of buyers.

Does this sound familiar?

Your SEO is solid. Your Google rankings are good. Your referral network produces steady work. But new leads are arriving through a channel you never optimized for. They say things like "ChatGPT recommended you." They book calls without comparing three other agencies first. They show up ready to engage because an AI model already told them you were the right fit.

Shaun Davidson noticed this pattern at Sembit and did what most agencies would not. He tracked it. Methodically. Over nine months. What he found was that 28% of their revenue originated from prospects whose buying journey started with a question typed into an LLM. Not Google. Not a referral. Not a conference. An AI chatbot.

That number would have been zero if Sembit had been invisible to AI models. Most agencies are invisible. They have never tested whether their name appears in AI-generated recommendations. They have never asked the question their ideal client would ask ChatGPT. They have no idea what the AI says about them, if it says anything at all.

Shaun did not just discover the problem. He built a methodology to measure it and a system to fix it. The approach is straightforward, repeatable, and available to any agency willing to invest the time.

Take the free Signal Check -->

Meet the Guest

Shaun Davidson is the VP at Sembit, a digital marketing agency, and Co-Founder of Zero Channel, a platform focused on helping businesses optimize their visibility in AI-generated recommendations.

Over 9 months, he tracked and validated that 28% of Sembit's revenue originated from prospects who found them through LLM recommendations. That discovery led him to develop a citation-first methodology for testing and improving AI visibility. He now helps other agencies understand whether they are visible to AI models and what to do about it.

Shaun's approach is data-driven and practical. He does not sell hype about the future of AI search. He measures what is happening right now and builds systems to capitalize on it.

Visit sembit.com and zerochannel.ai -->

Episode Summary

1. The 28% discovery

Shaun started noticing a pattern in how new clients were finding Sembit. Prospects were booking calls and mentioning, unprompted, that ChatGPT recommended the agency. Not that they found Sembit through Google and then used ChatGPT to validate the choice. ChatGPT was the starting point. The discovery channel.

He began tracking these leads specifically. Over nine months, he measured the revenue from prospects whose buying journey started with an LLM recommendation. The number was 28% of total revenue.

This was not a marketing experiment. Sembit did not set out to optimize for AI visibility. The discovery was organic. Shaun noticed the pattern, measured it, and then reverse-engineered what was driving it. The agencies appearing in AI recommendations shared specific characteristics that Shaun then codified into a methodology.

The implications are significant. A channel that no one on the marketing team was managing was producing more than a quarter of the company's revenue. For most agencies, this channel does not exist because they are invisible to it.

2. The citation-first methodology

Shaun developed a systematic approach to understanding how LLMs reference agencies. The method starts with identifying the prompts your ideal client would type into ChatGPT, Claude, or similar tools.

What would a mid-market ecommerce founder ask when looking for a digital marketing agency? What would a SaaS CEO type when searching for a development partner? These are the prompts that matter. Not generic queries about what agencies do, but specific, high-intent questions from buyers who are ready to hire.

Once you have your prompts, run them 100+ times. Document which agencies appear. Note how they are described. Track what sources the model cites. Look for patterns in which agencies appear consistently and which never show up.

This gives you a baseline. Without it, you are optimizing blind. You do not know whether you are visible, invisible, or mentioned but described poorly. The baseline tells you exactly where you stand and what needs to change.

"A prospect spent 30 minutes with ChatGPT, got our recommendation, searched us on Google, and booked. No comparison shopping."

3. Why directory profiles matter more than you think

Clutch, DesignRush, GoodFirms, and similar directory profiles carry outsized weight in AI model training data. These directories are structured, regularly updated, and densely linked. LLMs draw heavily from them when generating agency recommendations.

Shaun says agencies that ignore their directory profiles are leaving one of the easiest AI visibility levers on the table. The profiles are often free or low-cost. The information is within the agency's control. And the impact on AI recommendations is disproportionate to the effort required.

Complete profiles with reviews, case studies, accurate service descriptions, and up-to-date contact information improve your chances of appearing in AI-generated recommendations. Incomplete profiles, or profiles with outdated information, reduce your chances. The directories are effectively training data for AI models. Treating them as a low-priority marketing task is a mistake.

4. High-intent prompts over broad queries

Not all prompts are equal. Shaun focuses on high-intent queries. These are the specific questions a prospect asks when they are close to a hiring decision.

"Best Shopify agency for mid-market DTC brands" is more valuable than "what is a digital marketing agency." The first prompt comes from someone ready to hire. The second comes from someone doing general research. Optimizing for the first type concentrates your effort where it produces revenue.

Shaun recommends building a library of high-intent prompts specific to your agency's specialty, target market, and geographic focus. Then test each one repeatedly to understand where you stand. The prompts that matter most are the ones your ideal client would actually type when they are ready to make a decision.

5. The 5-prompt test

Shaun offers an actionable test any agency can run in 10 minutes. Write the prompt your ideal client would type into ChatGPT. Something like "recommend a [your specialty] agency for [your target client type]."

Run it 5 times. The results will vary because LLMs generate different outputs each time. If your agency appears in most runs, you have visibility. If your agency never appears, you have found the gap. If a competitor appears consistently and you do not, study why. Look at their directory profiles, their content footprint, and their online presence.

This simple test tells you whether you are visible or invisible in what may be the fastest-growing discovery channel for professional services. It takes 10 minutes. The information it provides is worth far more than the time invested.

6. What this means for the buying journey

The traditional agency buying journey started with Google, referrals, or industry events. A prospect identified a need, searched for options, compared agencies, and made a decision. Each step involved multiple touchpoints and significant time.

A growing share of buyers now starts with AI. They ask ChatGPT to recommend agencies. They get a shortlist. They research only the agencies on that list. The comparison shopping phase collapses because the AI already filtered the options. The prospect arrives at the agency's website with a recommendation in hand.

Shaun's example captures this perfectly: a prospect spent 30 minutes with ChatGPT, received Sembit's recommendation, searched them on Google to verify they were real, and booked a call. No comparison shopping. No evaluating five agencies. The AI did the shortlisting.

If you are not on the AI-generated shortlist, you never enter the consideration set. This is not replacing Google or referrals. It is adding a new layer before them. And most agencies have done nothing to show up in it.

Notable Quotes

"A prospect spent 30 minutes with ChatGPT, got our recommendation, searched us on Google, and booked. No comparison shopping."

"Run the prompt your ideal client would use. Five times. If you never appear, that's the gap."

"Directory profiles are not just for SEO anymore. They are training data for AI models."

"Twenty-eight percent of our revenue came from a channel nobody on the team was managing."

Related Episodes

If this one resonated, you'll like these:

Ep 10: Corey Quinn on specialization and why narrow focus makes you findable in every channel, including AI.
Ep 54: Jacob Miller on new client acquisition channels and adapting to how buyers actually find agencies.
Ep 15: Eddie Saunders Jr on visibility, consistency, and showing up where your buyers are looking.