FREE AI MARKETING TRAINING

Cheat Sheet

Updated on February 12, 2026

ChatGPT Projects for Marketing Operations

Use Projects to keep marketing work organized and grounded, so the AI produces consistently useful output instead of restarting, drifting generic, or going off the rails.

What you'll be able to do

  • Set up a Marketing Project that applies shared instructions to every chat inside it
  • Reduce cold start (starting from zero each session), boiled chicken (generic, bland output), and drunk uncle (hallucinations, forgetting, going off-track)
  • Attach core business context (offer + persona + plan) so output becomes specific and actionable
  • Use GPTs as tools inside a Project (e.g., writing voice or calendar builder) without losing context
  • Format instructions in Markdown so the AI reads structure (headings, bullets, tables) correctly

Key terms (quick definitions)

  • Growth OS: a proven system that helps founders and marketers plan → build → ship business growth campaigns with astonishing speed and quality
  • ChatGPT Project: a workspace that holds chats plus shared instructions and shared files
  • Project instructions: rules the AI follows for every chat in that Project
  • Grounding: anchoring responses to a trusted source of truth (files, URLs, docs, systems)
  • GPT (Custom GPT): a specialized tool you can call into a conversation to do a specific job
  • Markdown: a lightweight formatting language (headings = #, bold = **word**) that preserves document structure when you paste into ChatGPT
  • Customer Journey Framework: the marketing framework installed in the Project instructions — it teaches the AI how to move prospects through each stage of your funnel
  • Mechanic vs Strategy work: execution tasks vs high-leverage decisions that affect revenue

The 5 building blocks of ChatGPT

These five components work together — mastering how they interact is the real skill gap.

  1. Projects — workspaces that hold chats + instructions + files
  2. GPTs — specialized tools you call into a conversation
  3. Models — the AI engine (GPT-4o, o1, etc.)
  4. Threads — individual conversations inside a Project
  5. Tools — capabilities like web search, code interpreter, image gen

The 3 AI problems this setup solves

  • Cold start: every session feels like starting over
  • Boiled chicken: output is generic and "technically correct" but not useful
  • Drunk uncle: hallucinations, forgetting, and going off-track
Rule of thumb: the cure is Project structure + grounding (instructions + files).

Ways to ground your AI

Method What it does When to use it
Project instructions Rules the AI follows in every thread (e.g., customer journey framework) Always — this is your foundation
Project files Offer doc, persona doc, growth plan uploaded to the Project Always — minimum viable grounding
Web search Give ChatGPT a URL and it pulls live page content into the conversation When you need it to reference a specific page (landing page, competitor, article)
Thread attachments Upload a file directly into a single conversation When you need temporary context for one task
Stack them: Project instructions + Project files give you persistent grounding. Web search + thread attachments add one-off context on top.

Projects vs GPTs (how to think about it)

Projects GPTs (Custom GPTs)
Workspace / container for a category of work (e.g., Marketing) Tool you use for a specific task (e.g., rewrite, plan, analyze)
Holds chats + shared instructions + shared files Has its own instructions and (optional) knowledge
Creates compounding context across threads Pick it up, use it, then dismiss it
Best mental model: "Intelligent folder" Best mental model: "Hammer / screwdriver"

Do the work inside a Project, then call specialized GPTs when you need a specific capability—without losing your grounding.

Pro tip: use Markdown for Project instructions

When you paste rich text (bold, headings, tables) into ChatGPT, the formatting gets stripped. The AI loses the structure it needs to understand long instructions. Markdown fixes this.

  1. In Google Docs → Tools → Preferences → Enable Markdown
  2. Select your text → right-click → Copy as Markdown (or Edit → Copy as Markdown)
  3. Paste into your Project instructions — structure (headings, bullets, tables) is preserved
Bonus: You can also go to File → Download → Markdown (.md) to export an entire doc. An .md file is just a text file — nothing scary.

Build your Marketing Project (setup checklist)

  1. Create a new Project (name it "Marketing" or "Marketing Ops").
  2. Choose memory scope:
    • Project-only = sealed container (recommended; especially for client work)
    • Default = may reference info across your account
  3. Add Project instructions — paste the customer journey framework using Copy as Markdown so structure is preserved.
  4. Attach grounding files (minimum viable):
    • Offer (what you sell, who it's for, outcomes, proof)
    • Persona (pain points, triggers, objections, language)
  5. Do all marketing work inside this Project so context compounds over time (reuse and refine key threads).
Minimum viable grounding: Offer + Persona — even a single sentence like "I sell craft roasted coffee to busy corporate professionals" massively improves output. Best practice: Offer doc + Persona doc + Growth Plan.

When to use project-only memory

Project-only creates a sealed container — no memories leak in or out. Use it when:

  • Client work — keep Client A's context separate from Client B
  • Multiple businesses — prevent the AI from mixing up your offers
  • Personal vs professional — e.g., a nutrition/fitness project shouldn't bleed into your marketing project

Power tips from the training

  • Name your threads with a prefix so they're easy to scan (e.g., "Page: ClickPro landing page," "Promo: newsletter email")
  • Move existing chats into a Project — click the ⋯ menu on any chat and add it to a Project
  • Build personas fast — paste an example persona into the AI and say "build one like this for [my audience]." Brain-dump what you know; the AI fills in the gaps
  • Do 99% of your work inside a Project — working outside a Project means you're ungrounded

FAQ

What is a ChatGPT Project?
A Project is a workspace that groups related chats and can include shared instructions and files so every thread in that Project starts with the same context.
How is a Project different from a GPT?
A Project is the container where work lives (instructions + files + threads). A GPT is a specialized tool you call into a conversation to do a specific job (e.g., rewriting or building a calendar).
What does "grounding" mean?
Grounding means anchoring the AI to a trusted source of truth—like your offer doc, persona doc, plan, or a web page—so it produces relevant output and is less likely to hallucinate.
Should I use project-only memory?
Project-only is typically the safest default because it keeps context contained—especially important when working with clients, multiple businesses, or separate personal/professional areas.
Why does my output feel generic?
Generic output usually means the AI isn't grounded. Add project-level instructions (a framework) and attach files that define what you sell and who you sell to. Even a single sentence about your offer and persona makes a noticeable difference.
What should I attach to my ChatGPT Project first?
Start with (1) a short offer doc and (2) a persona doc. If you have one, add (3) a growth plan or marketing plan to increase specificity and strategic alignment. Markdown (.md) files are ideal, but PDFs work fine too.
How do I get a Word doc into Markdown?
The easiest route: paste your Word content into Google Docs, enable Markdown (Tools → Preferences → Enable Markdown), then use Edit → Copy as Markdown. You can also download the entire doc as a .md file via File → Download → Markdown.
Can I move an existing chat into a Project?
Yes. Click the ⋯ (ellipsis) menu on any chat in the sidebar and select the Project you want to add it to. The chat and its history will move into that Project.
How do I build a persona quickly?
Use an example persona as a model: paste it into the AI and say "build one exactly like this, but for [your audience]." Brain-dump everything you know about your customer, and the AI will organize it into the same structure.