PROGRAM SETUP
4 steps. ~5 minutes. Do this before building your CCD.
Step 1: Create your Claude Pro account ($20/month)
Go to claude.ai. Create an account or log in.
Settings → Subscription → Upgrade to Pro (important)
Step 2: Turn off data training
Settings → Privacy & Data → Toggle off “Help improve Claude.” If you haven’t done it yet, we suggest you do it before uploading anything.
Step 3: Create your Project
Sidebar → Projects → New Project. Name it: [Your Company] - Pipeline Build
A Project is a persistent workspace inside Claude. Files you upload stay available in every conversation inside it automatically. You never re-explain your business at the start of a session. You never re-paste your pitch deck. Claude already has it, because it’s all in the project.
Step 4: Follow the instructions in the CCD Toolkit module
WORKING WITH CLAUDE
The rules. 3 minutes. Bookmark it.
YOUR INPUT IS EVERYTHING
Claude writes 80%. You add the 20% that makes it real: your deals, your data, your judgment. That 20% is the edge. Without it, you have polished generic content. With it, you have a system that sounds like someone inside your business.
✅ Spend real time on your CCD.
It feeds all the modules. A thin CCD doesn’t break today’s output: it breaks everything downstream when every prompt references it.
✅ Include actual numbers: close rate, deal size, sales cycle length, conversion rates. Claude needs data to do math. “We’re growing” is not data.
✅ Be honest about what’s broken. Claude builds on whatever you provide, including an idealised version that isn’t true. If you tell Claude your close rate is 30% and it’s actually 8%, every pipeline target will be wrong.
❌ No generic input.
“We help businesses streamline operations.” This could be anyone. Claude will produce generic output from generic input.
❌ Say what’s actually different.
Vague differentiators: “great support,” “easy to use.” Every competitor says that.
⚠️ Garbage in, garbage out — but six weeks later. A weak input doesn’t break today’s output. It breaks everything downstream.
TECHNICAL RULES
• Cannot read URLs inside documents — paste links in the chat.
• Cannot log into your tools — it tells you what to do, you click.
• Knowledge has a cutoff — verify recent events and competitor claims.
• One file, one version — delete old before uploading new.
• Paste raw data — actual exports, not descriptions of data.
CONVERSATION HYGIENE
✅ Name files consistently. Station[X]-[Deliverable].
✅ Long threads dilute context. Start a new conversation for each major task.
✅ More context = smarter output. Keep your project files organized and up to date.
✅ One file, one version. When uploading a new version of a file, remove the old one.
✅ Files in Project Knowledge persist across all conversations vs Files attached in a single conversation stay only in that conversation
DATA PRIVACY
Data privacy: ✅ Investor-deck-level info. ❌ Never share passwords, API keys, or account numbers.
⚠️ Do not add sensitive information and data, or anonymize them.
ALWAYS VALIDATE
✅ Review every major output before using it.
You know your business better than Claude ever will.
✅ Verify every name, number, statistic, and competitor claim.
Claude can state wrong facts with full confidence.
✅ Push back when something doesn’t match reality.
Tell Claude what’s wrong and why. The next output will be dramatically better.
✅ Add your 20% to everything: real stories, proprietary data, personal experience.
❌ Skim and accept. If you didn’t validate it, you chose to let wrong assumptions spread.
❌ Use Claude’s output verbatim for anything external. Claude writes the structure. You add the soul.
WHEN OUTPUT ISN’T RIGHT
GENERIC OUTPUT
Could apply to any company
Fix: Your input is too vague. Add specificity and re-upload.
CONFIDENT ERRORS
Wrong facts stated with certainty
Fix: Verify all names, numbers, and claims before external use.
DOESN’T MATCH REALITY
Logical but your market disagrees
Fix: Tell Claude what’s wrong and why. Be specific. Regenerate.
Suggested Prompts
How can you ensure that the proprietary data and personal experiences you add truly enhance Claude’s output rather than introduce bias?
In what ways might inaccurate input data impact different stages of pipeline building, and how can you mitigate these effects over time?
How does organizing your projects and conversation threads strategically influence the efficiency and accuracy of collaborative work with Claude?
Module Comments 0 Analytics Settings
INTRODUCTION
MODULE
Introduction & Welcome
STATION 0: LAUNCH
MODULE
Program setup & working with Claude
MODULE
Company Context Document: Teach Claude Your Business
MODULE
CCD - Toolkit
STATION 1: ATTRACT
STATION 2: BUILD
STATION 3: ATTRACT
STATION 4: CAPTURE
STATION 5: AMPLIFY
STATION 6: MEASURE