Perpetual Traffic Podcast:

Episode – "The Little-Known AI Creative Research Tactic to Hack Meta Andromeda"

Date: January 13, 2026
Hosts: Ralph Burns & Lauren Petrullo
Guest: Cole Turner (Growth Strategist, Tier 11)

Episode Overview

This episode unveils a groundbreaking, actionable approach to market research using AI, specifically a "multistep prompt" process that leverages ChatGPT and similar LLMs for creative research—vital for succeeding on Meta’s Andromeda algorithm and other paid ad platforms in 2026. Returning guest Cole Turner walks through the methodology and live demos how teams at Tier 11 have collapsed weeks of time-consuming, manual research into a rapid, repeatable process that unlocks hyper-specific, emotionally resonant messaging to improve ad results.

Key Discussion Points & Insights

1. The Enduring Importance of Deep Market Research (06:14–12:12)

• Ralph Burns frames the discussion, recalling the evolution from slow, manual creative research to today’s AI acceleration:

  “One of the things that never goes out of style is good old-fashioned market research... In 2026, you still need to do the research ahead of time. Now, it’s just a less manual process.” (03:24)

• Cole Turner emphasizes why surface-level messaging underperforms and illustrates with a personal story:

  • Example: Two types of knee brace ads, one addressing surface pain and the other pressing on deep, identity-based emotions.

  “The deeper you can go down that iceberg, the more you’re going to be able to elicit emotion from someone... the way to get someone to take immediate action.” (10:36)

• Lauren Petrullo reframes specificity as finding each person’s “flavor of ice cream” instead of serving vanilla to everyone (11:00–12:10).

2. Messaging Diversification & Testing for Campaign Resonance (12:12–15:41)

• The group debates channel-specific messaging: Should all channels reinforce the same angle, or do you diversify and iterate to see what resonates?
• Cole Turner references Eugene Schwartz’s “levels of awareness” model as essential to smart creative segmentation:

  “There’s different stages... you’d want to touch someone at a problem-aware stage with cold channels... But on Google Search, it’s more, ‘We know you have this problem, this has worked for others, buy this.’” (13:49)

3. The AI-Powered Market Research Tactic: Step-by-Step Walkthrough (15:41–24:17)

The Old Way vs. The New Way

• Previously: Manual scraping of Facebook, Reddit, long hours for each market.
• Now: Two-step AI process—fast, scalable, and more comprehensive.

Step 1: The “Prompt-for-a-Prompt” Technique

• Input prompt instructs ChatGPT to “take on the role” of a seasoned market researcher and direct response copywriter (David Ogilvy, Eugene Schwartz).
• Tells it to build a robust research plan for a problem-aware segment (illustrated with “people with lower back pain who could benefit from a standing desk”).

Step 2: Automated Internet-Wide Research

• Output: The first AI prompt builds a detailed secondary-prompt for “deep research.”
• This is pasted into ChatGPT (or Gemini/Claude with web access). AI scours Reddit, blogs, and forums for authentic “voice of customer” quotes, problems, language, and emotional triggers.
• The final output is in-depth, with flagged quotes ready for copy, hooks, and personas.

Cole Turner (20:44):

“This will take us ages to go through—a huge research dossier... like, ‘my lower back is screaming.’ That’s a quote from Reddit. If you put this in an ad, someone is going to read that and say, ‘Wow, they really get me…’”

4. Real-Time AI Copy Generation: Demo & Discussion (24:17–27:37)

• Cole uploads the research base to a custom GPT (20 mins from scratch) and demos real-time copy creation:

  • Prompt: “Write me a YouTube ad script using the attached market research to sell a standing desk. Keep it around 60 seconds, use the problem–agitate–solution framework. Use an obvious yes question as a hook.”

  • Sample AI Output:
    "Quick question: Does your lower back start screaming by mid-afternoon after sitting all day?”

  • This is immediately lauded as more powerful and specific than generic copy.

• Lauren Petrullo:

  “You make me want to use my standing desk right now… it’s the middle of the day… I don’t want to have lower back pain!” (25:38)

• The group assures listeners: This isn’t meant to replace talented creatives, but it makes research and starting drafts far richer and faster than ever.

5. Prompt Layering & the Art of Prompt Engineering (27:44–34:45)

• Best practice: Use multi-step prompts especially when you lack deep domain expertise.
• Let the AI generate its own secondary instructions based on your needs and its strengths.
• If your prompt is lazy (“half-assed”), your output will be basic; treat LLMs as more than a search engine.
• Always assign the LLM an expert persona role (e.g., “25-year fitness coach for women”) to improve specificity.

Cole Turner (32:26):

“Your output from any LLM is only going to be as good as the input... Otherwise, garbage in is garbage out.”
— Echoed by Lauren Petrullo (29:36): “Garbage in, garbage out.”

• Lauren shares her approach to prompts: iterative, exploratory, with back-and-forth clarification for a much better research base (29:39–30:27).

6. Application, Guardrails, and Key Watchouts (35:19–36:39)

• This research is the essential base for “creative diversification,” maximizing Andromeda’s new features and reaching the right segments.
• The process is freely available (no email required):
  • Download the master prompt at tier11.com/prompt
• Crucial caveats:
  • Don’t blindly trust all AI output.
  • Review, tweak, and cross-check for correctness and fit.
  • Use this as a starting point—human judgment is still irreplaceable.

Cole Turner (36:07):

“LLMs aren’t quite good enough to replace a lot of roles... Don’t just blindly copy and paste this and accept every output that it gives you. Read through it, make sure it looks correct. If it needs to be tweaked, go ahead and tweak it.”

Notable Quotes & Memorable Moments

The Ice Cream Analogy for Messaging Specificity

Lauren Petrullo (11:00):

“When you’re talking to everyone, no one’s actually paying attention... You get to meet someone at their particular flavor of gelato. In this case it’s knee pain!”

"Prompt-for-a-Prompt" as a New Paradigm

Cole Turner (16:52):

“The first prompt we’re going to put into ChatGPT is actually going to give us back another prompt, that we’re later going to give to ChatGPT again... We’re telling it to create a research plan, a huge outline on that market...”

The Power of Authentic Research

Cole Turner (21:21):

“Voice of customer phrase bank... ‘My lower back is screaming’… that alone is a hook on an ad.”

Practicality Comes First

Cole Turner (35:55):

“Read through it, make sure it looks correct. If it needs to be tweaked, go ahead and tweak it. Don’t just blindly copy and paste this…”

Important Timestamps & Segments

• Market research’s timeless role – 03:24–07:10
• Personal/emotional depth in messaging – 07:10–12:12
• Awareness stages & channel deployment – 13:49–14:38
• Manual research vs. new AI process – 15:41–16:52
• The multistep AI research method: demo – 16:52–24:17
• Live copywriting prompt in action – 24:12–27:37
• Prompt engineering best practices – 27:44–34:45
• Final tips/cautions for AI market research – 35:19–36:39

Resources

• Download the master AI research prompt (no opt-in required)

Closing Thoughts

This episode gives listeners an exact, replicable research shortcut to generate deep, voice-of-customer insights for creative testing—absolutely critical for nailing Meta’s Andromeda and modern performance marketing. The key: thoughtful prompt engineering, AI as an accelerant (not a replacement), and always ensuring the output is as humanly relevant as possible.

Final wisdom:
Don’t skip the research. Use AI to do it smarter, faster, and with more emotional resonance—but always, always check before you trust.