Playbook · ChatGPT
ChatGPT for marketing: the workflows that hold up
Seven workflows, the setup that makes them good, and the four jobs to never hand over.
- 7 workflows
- Free tier viable
- What not to trust
ChatGPT for marketing: the workflows that hold up: what should you actually do?
The marketing jobs ChatGPT does reliably are drafting, repurposing, summarising research you supply, generating variants and analysing data you upload. It is unreliable for facts, statistics, competitor pricing and anything requiring judgement about your market. A context file plus a verification rule turns generic output into usable output; nothing else matters as much.
Last updated 3 August 2026
| Play | Effort | Cost |
|---|---|---|
| Set up a project with a context file | 90 minutes, once | Free |
| Turn customer emails into content briefs | 1 hour | Free |
| Repurpose one strong piece into four formats | 45 minutes | Free |
| Draft emails, then cut them hard | 2 hours | Free |
| Analyse your own exports weekly | 45 minutes a week | Free tier limits apply; Plus is $20/month |
| Stress-test your positioning before you publish it | 45 minutes | Free |
| Write a house rule for facts and data | 30 minutes, once | Free |
Most people get generic marketing output from ChatGPT because they give it a generic brief. The model has no idea what you sell, who buys it, what you charge or what has already worked. Fixing that one thing changes the output more than any prompt technique, any custom instruction set, and any of the paid wrappers built on top of it.
The second thing that matters is knowing which jobs it fails at. It will produce a market size, a competitor's pricing tier or a conversion benchmark with total confidence and no source. Published unchecked, that is a credibility problem that costs more than the workflow saved.
What follows is the setup, then seven workflows that survive contact with real marketing work, then the jobs to keep away from it.
- — Founders and small teams doing their own marketing with limited hours
- — Marketers who have tried ChatGPT, got bland output, and want to know if the problem was theirs
- — Anyone deciding between a $20 general model and a $99 marketing-specific AI tool
- — Teams that need a written standard for how their people use AI
Regulated industries publishing claims that need legal review, or teams needing audit trails and data-residency guarantees. Those need enterprise tooling with a data-processing agreement, not a consumer subscription.
The moves
Ranked, highest return first.
Work down the list. Each one names the first step so there's nothing to plan.
- 01
Set up a project with a context file
Every other workflow on this list is two to three times better with this in place. The difference between useful and generic output is almost always missing context, not prompt wording.
First step: Write 600-800 words on what you sell, who buys, why, your three competitors, your prices, your tone, and your three best-performing marketing activities. Put it in a project so every chat starts with it.
- Tools
- ChatGPT projects, a text file
- Effort
- 90 minutes, once
- Cost
- Free
- 02
Turn customer emails into content briefs
Questions from real customers are the only topic source that guarantees demand. The model is good at clustering them into themes and spotting the ones you answer most often.
First step: Paste 30 recent customer or prospect emails, ask for the ten recurring questions ranked by frequency, and turn the top three into content briefs including which of your own numbers each should contain.
- Tools
- ChatGPT, your inbox
- Effort
- 1 hour
- Cost
- Free
- 03
Repurpose one strong piece into four formats
This is where the model is genuinely excellent: it keeps the substance while changing the shape. Creating four new pieces is where it is mediocre.
First step: Give it your best-performing piece and ask for a plain-text social post, an email, a short video script and a slide outline — instructing it to keep every specific number from the original.
- Tools
- ChatGPT
- Effort
- 45 minutes
- Cost
- Free
- 04
Draft emails, then cut them hard
AI email drafts are structurally fine and tonally wrong. The reliable fix is subtraction: what remains after removing empty sentences reads like a person wrote it in a hurry, which is what people respond to.
First step: Draft a five-email sequence, then delete every sentence without a fact, a number or a story. Expect to keep about 40%. Send that.
- Tools
- ChatGPT, your email platform
- Effort
- 2 hours
- Cost
- Free
- 05
Analyse your own exports weekly
Uploading real data is the highest-value use in this list because the model is reasoning about your numbers rather than generating text about a generic business.
First step: Export 90 days of analytics or sales as CSV, upload it, and ask for the three biggest changes with likely causes and what you would need to check to confirm each. Then check them.
- Tools
- ChatGPT (file upload), your analytics export
- Effort
- 45 minutes a week
- Cost
- Free tier limits apply; Plus is $20/month
- 06
Stress-test your positioning before you publish it
Asking a model to argue against you is cheaper than learning the objection from a lost deal. It is good at generating the obvious counterarguments you have stopped seeing.
First step: Paste your homepage copy and ask it to respond as your three hardest prospect types, listing the objection each has in the first ten seconds. Fix the two that appear for all three.
- Tools
- ChatGPT
- Effort
- 45 minutes
- Cost
- Free
- 07
Write a house rule for facts and data
Two mistakes cost real money: publishing an invented statistic and pasting customer data into a consumer account. A written rule stops both; good intentions do not.
First step: One paragraph: no customer personal data goes in, every number needs a source link opened by a human, and one named person approves anything public. Pin it where the team works.
- Tools
- A shared doc
- Effort
- 30 minutes, once
- Cost
- Free
Sequence
What to do first, week by week.
Context file and one workflow
Write the context file and run only the repurposing workflow. It gives a visible result in under an hour, which is the fastest way to see whether the context file is doing its job.
Content briefs and email
Move to work that publishes. Ship one piece of content and one email sequence, applying the subtraction edit both times. Compare engagement against your previous three sends.
Data analysis and positioning
Start the weekly export habit and run the positioning stress test. These change decisions rather than adding output, which is where the durable value sits.
Rules and review
Write the house rule and review hours saved against results. Keep the two workflows that earned their time and drop the rest rather than trying to maintain all seven.
Avoid
Where this usually goes wrong.
Prompting without context
'Write me a marketing email' produces the average of the internet. The same request with a context file produces something recognisably about your business. This is the whole difference.
Publishing statistics it produced
It generates plausible, precise, invented numbers. Every figure needs a source you opened. This is the single most common way small teams damage their credibility with AI.
Asking for strategy
It will give you a confident, generic strategy assembled from public marketing writing. It does not know your margins, your churn, or which customers are painful. Use it to pressure-test a decision you made, not to make one.
Accepting the first draft
First drafts are competent and forgettable. The editing pass where you add your own data, examples and opinions is not optional overhead — it is where the value is created.
Pasting customer data into a personal account
Consumer accounts are not covered by a data-processing agreement. Anonymise exports or use a business tier before uploading anything containing customer names or contact details.
Questions
Common questions.
Is ChatGPT good for marketing?
For drafting, repurposing, summarising and analysing data you supply, yes — it saves most small teams four to six hours a week. For facts, statistics and strategy it is unreliable, and treating it as a source rather than a drafting tool is where teams get burned.
Do I need ChatGPT Plus for marketing work?
Not immediately. The free tier covers drafting and repurposing. Plus at about $20 a month is worth it once you are uploading files weekly or hitting usage limits, which usually happens in the second month.
What are the best ChatGPT prompts for marketing?
Prompt wording matters far less than context. A mediocre prompt with a 700-word context file attached beats an elaborate prompt with none. Spend the time on the context file.
Will Google penalise content written with ChatGPT?
Google's guidance targets unhelpful content rather than how it was produced. Unedited output tends to be unhelpful because it contains nothing new, which is the actual problem. Edited work with your own data performs normally.
Should I use ChatGPT or a dedicated AI marketing tool?
Start with ChatGPT for a month and note which jobs you repeat. Buy a dedicated tool only when it holds context you keep re-pasting, connects to a channel and acts, or reads live data the model cannot see.
More playbooks
Other plans worth reading.
AI-powered content creation without publishing filler
ReadPillar · AI marketingAI marketing: the complete, honest guide
ReadPlaybook · B2B SaaSGrowth plan for a pre-revenue B2B SaaS
ReadPlaybook · EcommerceGrowth plan for an ecommerce store's first 100 orders
ReadPlaybook · LocalGrowth plan for a local service business
ReadPlaybook · B2B SaaSGrowth plan for a B2B SaaS with early traction
ReadElsewhere on gtm.help
Keep exploring.
Free marketing tools
Calculators and generators that run without a signup.
BrowseTool guides
What to use, what it costs, when it earns its keep.
BrowseTool comparisons
Head-to-head picks for the calls people get stuck on.
BrowseMarketing checklists
Task lists by business type and competitive situation.
BrowseStraight answers
The growth questions founders ask, answered with numbers.
BrowseGet started
Get this ranked for your business.
Paste your URL and you'll have a plan built from your site and competitors in about a minute.