Most customer research I've seen dies in a slide deck.

An insights team spends months and serious budget painting a picture of the audience. They present the readout, then it lives as a file on the server that occasionally gets dug up for big projects.

Here's the better move: turn that research into a customer panel you can consult anytime.

Talking to your customers > reading about them

This isn't exactly talking to your customers, but it's the next best thing.

OLD: sifting through static reports for a few nuggets to drive strategy
NEW: interacting with personas to see how customers respond before you act

Customer research used to shape the big vision, now it can help with your everyday calls too.

Working on a new offer, product angles, or about to ship copy? Run them past your personas first.

The people I run everything past

I have a panel of personas developed from deep research runs, call transcripts, social signals, and feedback gathered over time. They're a mix of stakeholders:

The marketing decision maker
Her CMO boss
The direct reports who are my end users
The finance and IT leads who often sign off on the purchase

I packaged them into a Claude Skill I use to get feedback from a single persona or the whole group. I have them react to whatever I'm working on: website copy, UI, pitch materials, and more. Then I get blunt reactions and ideas to make it better for my audience.

I'm constantly impressed by what they come back with. Instead of using AI to scan documents, having it react as my customers gets far more useful results.

My panel reacting to draft website copy.

Run this: create your own customer panel by loading up your research into an AI Skill or Project that you can pressure-test any copy, offer, or idea against in seconds.

Build your own customer panel in 4 steps

1. Gather your research

Pull together everything that creates a full picture of your audience. Use things like:

Latest consumer insights reports and research readouts
Survey and poll responses (especially the verbatim open-ends)
Discovery, sales, or user interview transcripts and notes
Win/loss notes and cancellation reasons
Support tickets, reviews, and social comments

2. Have AI synthesize it into personas

Paste this prompt into a new chat window and attach the files you gathered in step 1.

You're building a customer panel I'll pressure-test my marketing against. The files I've attached are your only source of truth.
 
Read everything first. Find the distinct customer segments in the research and build one persona per segment it genuinely supports. If the research points to a single core customer, build one. Don't invent segments that aren't there.
 
For each persona, write:
- Who they are and what they're trying to do
- Their situation and how much they know about my category
- What they're feeling, and why
- How they read and decide: what they look for first, what they skim, what earns or loses their trust
- The exact words and phrases they use (pull real language from the research)
- The words or claims they'd misread, and what they'd think instead
- What makes them say yes, and what makes them walk away
 
Then write the rules the panel follows:
- React only as these personas, grounded in the research
- Never invent an objection the research doesn't support. If it's not covered, say so
- Flag any word in my copy these customers wouldn't use themselves
- Call out the specific files I attached and point to where deeper detail lives in each
 
Then set the review format: when I paste in copy, an offer, or an idea, each persona gives their first impression, how they read it, what makes them hesitate, what's missing, and whether they'd act.
 
One more rule: if the research doesn't give you enough to build a grounded persona, stop and tell me exactly what's missing and what to add, instead of guessing.
 
Write all of this as one instruction set I can save and reuse.

3. Make your panel reusable

Take the instructions from step 2 and save them. This looks a little different on each platform:

Claude: Tell Claude to create a Skill that uses the instructions and attaches your research files for reference. Name the skill '/customer-panel'.
ChatGPT: Create a new Project named 'Customer Panel.' Attach the instructions and upload your research files.
Copilot or Gemini: Create a new Notebook named 'Customer Panel.' Attach the instructions and upload your research files.

New to Skills, Projects, or Notebooks? Interactive lessons that get you started →

4. Get immediate feedback

Anytime you have a question, offer, copy, or a new campaign you want a read on, run it past them by doing the following:

Claude: Type '/customer-panel' in any chat window
ChatGPT: Open your 'Customer Panel' Project and start a fresh chat
Copilot or Gemini: Open your 'Customer Panel' Notebook and start a fresh chat

That's the move and it's easy for the value it returns.

Your insights work goes further and your customers get a say in everything you ship.

Learn AI by doing.

Set Up Your First Claude Skill in <15 Minutes

Learn, step by step, how to use Claude’s Skills feature to save time and deliver consistent results.

Start this module

Set Up Your First ChatGPT Project in <15 Minutes

Go from one-off chats to using AI strategically by creating a dedicated space that holds your files, your past work, and how you want ChatGPT to respond.

Start this module

See more modules →