21 January 2025

This AI Prompt Gets You Customer Insights in 5 Minutes (Free Tool)

2Frameworks

10Insights

Frameworks in this episode

Marketing6 steps

Recorded-Customer Digital Twin

Turn interview transcripts into a queryable voice-of-customer panel

Read →

Marketing6 steps

Synthetic Persona Focus Group

Simulate diverse buyers, collect individual reactions, then synthesize the verdict

Read →

Insights & moments

The myth-busts, hot takes, explainers, and tools worth keeping.

Hot Take· 2

Hot Take12:00

Why YouTube and Podcast Transcripts Can Produce Fresher Content

Kieran argues that public text is often already represented in model training data, reducing its value as differentiated source material. He recommends grounding content creation in video and podcast transcripts, which can supply less commoditized ideas for text-based outputs.

Common web text may already be reflected in model outputs
Video and podcast material can offer differentiated source context
NotebookLM can ingest multiple YouTube sources
Transcripts can be transformed into articles or social posts

“the Treasure Trove right now for Content creators and how they use AI is data that is non-text”
Kieran Flanagan · 12:00

“it doesn't have like YouTube or it doesn't have podcasts”
Kieran Flanagan · 12:30

#content repurposing#youtube#podcasts#training data

Hot Take25:30

Google Search May Become a Task-Routing AI Interface

Kieran predicts that users will stop selecting individual models or research modes themselves. Instead, Google may infer the task and automatically route it to blue links, deep research, Gemini, or another specialized process.

Different models and interfaces currently suit different tasks
Agent systems already route work among specialized components
Future search may choose the appropriate process automatically
Users may care about the result rather than the underlying model

“you won't have to worry about that because it will just orchestrate all of that thing for you”
Kieran Flanagan · 26:30

“Google will actually decide if it's going to give you the Blue Links if it's going to run deep research if it's going to run…”
Kieran Flanagan · 26:30

#google search#ai agents#orchestration#future of search

Explainer· 3

Explainer00:30

Why AI Changes the Economics of Market Research

Traditional in-depth focus groups can cost thousands of dollars and take weeks or months to complete. The hosts argue that AI can compress early directional feedback into minutes, allowing teams to explore far more questions before paying for formal validation.

Traditional focus groups can cost roughly $8,000 to $12,000
Conventional research may take weeks or months
AI makes early feedback dramatically faster and cheaper
Speed enables teams to test more possible directions

“what would normally cost the average company probably eight to 12,000”
Mike Taylor · 00:30

“it's taking weeks and months and condensing that same feedback to minutes for for free versus that 8 to 12 Grand”
Kieran Flanagan · 01:00

#market research#cost reduction#speed#ai

Explainer15:30

Move From One Campaign Message to Individual Positioning

The hosts extend message research into personalization: once customer context exists, AI can identify the most relevant positioning angle for each person or audience cluster. Taylor recommends using the system to discover an angle of attack and then exercising human judgment over the final writing.

A top-level message can be adapted for individual customers
Customer transcripts reveal person-specific motivations
Audience clusters provide a scalable middle ground
Generated copy can be used for insight rather than sent verbatim

“what is the best single best way to position it for each individual person or at minimum like group of people”
Kipp Bodnar · 16:00

“once I understand the angle of attack you know the approach then then I'll kind of write it myself quite often”
Mike Taylor · 16:30

#personalization#email marketing#positioning#customer data

Explainer21:30

How Internet Training Data Reveals Persistent Customer Problems

Taylor explains that models can identify familiar problems such as cash flow because their training data contains discussions from forums, support communities, and other public sources. Repeated associations allow the model to infer common concerns among specific customer groups.

Training data acts like a broad snapshot of public internet discussion
Niche forums contain detailed expressions of customer pain
Repeated patterns connect customer types with recurring problems
These inferences can help generate demand hypotheses

“you can kind of think of the training data as like just a big snapshot of the internet”
Mike Taylor · 21:30

“small business owners are always talking about cash flow issues”
Mike Taylor · 22:00

#training data#customer pain#small business#demand research

Story· 1

Story19:00

The Vertical AI App Opportunity Around Latent Demand

The hosts propose building narrowly focused applications for markets such as Amazon sellers, Shopify merchants, education businesses, or fitness. Such an app could identify underserved problems, suggest products, develop positioning, and support rapid business creation.

Vertical applications can encode market-specific context
Synthetic buyers can surface underserved problems
The workflow can extend from demand discovery to product positioning
Narrow market expertise can make the application more useful

“I would make a bunch of very verticalized apps for for product or business creation”
Kipp Bodnar · 19:00

“what is missing in the market in terms of a product that has latent demand”
Kipp Bodnar · 19:30

#vertical ai#app ideas#latent demand#entrepreneurship

Tool· 2

Tool07:30

Measuring Brand Visibility Inside AI Training Data

The hosts describe an AI search visibility tool that compares how frequently a brand appears in relation to a business category versus its competitors. They suggest this can provide a rapid directional alternative to some traditional awareness studies.

AI systems encode associations between brands and categories
Brand mentions can be compared with competitors
The hosts report results close to existing HubSpot data
The method offers faster directional awareness measurement

“it shows you how visible you are in the training set versus other brands”
Kieran Flanagan · 08:00

“it's within a margin of error of the”
Kipp Bodnar · 08:00

#brand awareness#ai search#visibility#competitive analysis

Tool24:00

Google Deep Research Turns Hundreds of Pages Into One Analysis

Kieran describes Google Deep Research as an automated research process that creates a plan, traverses many websites, explains its methodology, and produces structured outputs. The example research on focus-group spending included cost components and a table exportable to Sheets.

The tool develops and executes a research plan
It can inspect dozens or hundreds of websites
Outputs may include methodology, calculations, and structured tables
Automated synthesis replaces extensive manual search traversal

“I'm researching for 27 websites now I'm researching for 41 websites now I'm looking over like a 100 websites to pull together all of the…”
Kieran Flanagan · 24:30

“it went through 29 different websites told me its methodology but had a sick table that I could then export to sheets”
Kieran Flanagan · 25:00

#google deep research#web research#automation#gemini

Takeaway· 2

Takeaway14:00

Stock AI Answers Will Not Differentiate Your Content

Taylor warns that faster generation alone will cease to be an advantage when everyone uses the same models. Proprietary interviews and internal data create a distinct source of insight that competitors cannot reproduce from a stock prompt.

Generic AI answers are available to every competitor
Generation speed is becoming commoditized
First-party interviews create proprietary context
Differentiation depends on supplying information others lack

“having an answer that other people don't have”
Mike Taylor · 14:00

“if you're just using the stock answer from chat gbt or from Claude uh you don't have any real differentiator over anyone else who's using…”
Mike Taylor · 14:00

#differentiation#first-party data#content strategy#competitive advantage

Takeaway27:00

If AI Is Not in Your Workflow, Start With Research

The episode recommends research as the easiest and least controversial entry point for practical AI adoption. Teams can use it to answer low-cost directional questions, reserve human research budgets for the highest-value uncertainties, and validate the final conclusions conventionally.

Research requires almost no setup to begin
AI can answer questions that budgets would otherwise leave unexplored
Directional findings can still be validated by humans
Research use avoids some objections associated with publishing generated media

“just use an AI for research”
Kieran Flanagan · 27:00

“if you can use AI to answer those other 99 questions you otherwise wouldn't have been able to ask um then you can really direct…”
Mike Taylor · 29:30

#ai adoption#research#workflow#validation