# How to Use the Messaging Discovery Guide
### A toolkit from The Goods

## What you're getting

A structured conversation with an AI, set up to think the way Cerys does when she works with founders on messaging: it won't let you skip to "what should we say" before the business and customer thinking underneath it is solid. It will push back on generic answers, ask for proof, and tell you plainly where you have gaps — not just hand you a tagline.

You can do this in one sitting or spread it across a few days. You can type, or you can talk. By the end, you'll have a one-page message architecture and a clear list of what to go build or decide before you say it publicly with full confidence.

This is designed to be self-serve. If you get stuck, or you're about to finalize messaging ahead of something high-stakes (a raise, a launch, a big partner conversation), the tool itself will tell you when it's time to bring Cerys in — you don't have to guess.

## What you need

- An AI tool that supports **Projects** (or an equivalent persistent workspace) — Claude and ChatGPT both work well
- About 60–90 minutes of real thinking time, though you don't need to do it all at once
- Nothing else. No deck, no prior messaging doc required — the tool will surface what it needs from you

## Setup (takes about 5 minutes)

1. **Download the toolkit file** — `messaging-toolkit-ai-instructions.md` — from this page.
2. **Open your AI tool and start a new Project.**
   - In Claude: go to Projects in the sidebar, click New Project, give it a name like "Messaging Discovery."
   - In ChatGPT: create a new Project from the Projects area, or set up a Custom GPT if you prefer.
3. **Find the custom/project instructions field** and paste in everything from the toolkit file below the line marked "PASTE EVERYTHING BELOW THIS LINE." You don't need the file's own header — just the instructions themselves.
4. **Save it.** Every conversation you start inside this project will now follow the process automatically — you don't need to re-explain it each time.
5. **Start a new chat inside the project** and say something like: *"I'm ready to start."* It will take it from there.

## Two ways to work through it: type or talk

**Typing** works exactly the way you'd expect — answer its questions in the chat box, stage by stage.

**Talking** works too, in a few different ways depending on your tool:

- Most AI apps have a **voice mode** built in (usually a microphone icon near the chat box) that lets you speak and get a spoken or written response back in real time — good for thinking out loud through a stage.
- Mobile apps often support **dictation** — tap the mic, speak, and your words appear as editable text before you send them, so you can clean it up first if you want.
- If you'd rather record separately (a voice memo on your phone, a call with a cofounder), use any transcription tool to turn it into text, then paste the transcript into the chat. Tell it plainly that it's a transcript — the tool is built to handle rambling, unstructured spoken answers and pull out what it needs.

None of these need to be perfect. Talking through a messy, half-formed answer is often exactly how the real thinking surfaces — the tool's job is to sharpen it from there, not to expect a clean answer up front.

## What to expect while you're in it

- It will ask before it accepts. If your answer is vague or sounds like it could apply to any competitor, it will say so and ask again — that's the process working, not a bug.
- It will separate what you know from what you believe. Anything you can't back up with evidence gets marked as an assumption, not folded into your message as fact.
- It will occasionally tell you to stop and go build something — a case study, a data point, a customer conversation — before continuing. That's the sequencing doing its job: the message can only be as strong as what's underneath it.
- You can ask it to "show progress" at any point to see the running log of what's validated, what's assumed, and what's still open.

## When to loop in Cerys

The tool is built to flag this itself, but as a rule of thumb, reach out (cerys@thegoods.io) if:

- You've hit the same gate twice and genuinely can't move past it
- You're finalizing messaging ahead of a fundraise, launch, or major partner conversation
- The tool surfaces a business-model or customer-definition gap that feels bigger than a chat can resolve
- You just want a second, human read before you commit to it

## What you'll walk away with

- A one-page **Message Architecture**: your umbrella message plus three supporting messages, each tagged by how solid the evidence behind it is
- A **Gap Assessment**: the business decisions still open, and the specific proof points you need to go build, in priority order
- A clear sense of which parts of your current pitch are genuinely yours — and which parts you were borrowing from the category without realizing it