How We Used Deep Research to Build Colby Davis’s Content Library

During Office Hours this week, we broke down how to use AI the right way—by doing real research, not generating fluff.

Our test subject: Colby Davis, CEO of Davis Painting, one of the largest residential painting companies in the U.S. Despite running a major company, Colby had very little personal brand presence online. No strong Google profile. No articles about him (outside of their own website or ones we’ve published). Just his company site.

This made him the perfect case study for testing our Content Library creation process using AI tools plus human review.

Why Colby Davis?

• Runs a multimillion-dollar painting business in New Jersey.
• Has expanded to five states through acquisitions
• Featured in a Young Entrepreneur Council article, but not indexed well
• No knowledge panel
• No rich snippets or third-party citations

In short – we needed to build his E-E-A-T and positive mention catalog from scratch.

Step 1: Asking AI Without Context = Weak Output

We tested 3 tools:

• ChatGPT
• Grok, via Chad “CBT” Brock
• Gemini (Google’s new model)

Prompt:

“Create a comprehensive profile of Colby Davis, CEO of Davis Painting.”

The results?

All three models struggled:

• Confused Colby Davis with a jewelry brand in Boston.
• Mistook Davis Painting for Davis College.
• Pulled from DavisPainting.com—but nothing else.
• Cited few sources and didn’t distinguish real Colby from noise

This is what low EEAT looks like:

• No stories or interviews = no Experience
• No how-to or third-party citations = no Expertise or Authority
• No testimonials or case studies = no Trust

Step 2: Activating Deep Research Mode

Deep research, when using AI tools like ChatGPT or Grok, refers to the process of guiding the model to explore a topic thoroughly, thoughtfully, and across multiple dimensions.

Rather than stopping at surface-level summaries or quick facts, deep research involves asking the AI to analyze causes, implications, relationships, and competing perspectives.

It includes breaking the topic into sub-questions, examining historical context, comparing expert viewpoints, and looking at real-world examples or case studies.

The goal is to develop a layered understanding that mimics how a skilled researcher might explore a subject through reading, critical thinking, and synthesis.

We re-ran the same prompt, this time guiding each tool with context from our SOPs and source links.

Here’s what changed:

ChatGPT:

• Pulled from Dennis’s previous blog post on service businesses.
• Grabbed snippets from Colby’s LinkedIn.
• Organized results into a 4-part structure: background, business model, marketing, growth.

 Grok:

• Indexed 72 sources in 54 seconds.
• Surfaced social mentions from Facebook and Nextdoor.
• Found video clips Colby appeared in (e.g., a 2022 home improvement expo in Austin).
• Merged timeline and topic clusters.

Gemini:

• Filtered out false positives.
• Noted that the YEC article existed but wasn’t prominent on Google.
• Suggested DavisPainting.com pages that should have schema markup but didn’t.

Step 3: Comparing AI Output

We compared side-by-side results.

All 3 needed editing. None nailed the voice or narrative without human help.

Step 4: Topic Wheel Test — and Why AI Failed

We prompted each AI to build a Topic Wheel using our core framework:

1 Hub topic + 6 spokes = Topic Wheel.

For example, this is mine:

AI had a few suggestions for Colby’s:

• Entrepreneurship.
• Business Ethics.
• Marketing.
• Home Improvement.
• Customer Service.
• Leadership.

Sounds fine—but generic. Not specific to Colby’s real journey.

What was missing?

• “How to buy local competitors in low-trust markets”
• “What to watch for when expanding painting crews across states”
• “Why Nextdoor and Facebook Reviews are worth more than Yelp in home services”

That’s where the human editor had to step in.

Step 5: Adding the Human Layer

We cleaned up the content and added real EEAT elements:

✅ Colby’s business stats (5 states, 200+ employees)
✅ His acquisition strategy (targeting companies with <3 Google reviews)
✅ Quotes from his team on training and retention
✅ His actual marketing stack (ServiceTitan, Birdeye, Facebook Reviews)

We also linked out to relevant BlitzMetrics resources, including:

• How to Build a Topic Wheel
• Why You Need a Positive Mentions Strategy
• How E-E-A-T Impacts Your Content

Step 6: Final Output — A Real Content Library

Now that we’ve built a base, Colby’s team can:

• Expand the Topic Wheel into blogs and videos.
• Trigger a Google Knowledge Panel.
• Rank for name, brand, and niche terms.
• Use the LDT Framework to show proof over time.

As AI tools get smarter the process of collecting positive mentions and building a content library will get easier. This month we’ve launched Ethan, our Positive Mention helper who’s job is to collect positive mentions wherever he finds them online.

AI can get 80% of the way. But only you—or a well-trained human—can take it to the finish line, provided there’s enough existing EEAT content available which search engines can tap into.