Most people use AI assistants for content the way some people use a kitchen: walk in hungry, open every cupboard and make whatever seems possible. Sometimes it’s dinner. Sometimes it’s toast for the third night running.
Professional kitchens don’t work that way. They work in stations. Each station receives something specific and passes something specific along. The output is predictable not because the cooks are robots, but because the workflow is.
This article is a teardown of a content workflow built the same way: ten stages, each with its own set of prompts, each producing an input for the next. It follows how the SEO & Content Prompt Pack organizes its 120 prompts, but the structure is useful whichever tools you use.
Why organize prompts by workflow stage?
Because the hardest question in AI-assisted content isn’t “how do I phrase this?” It’s “what should I be doing right now?” A library organized by stage answers that before you ask it.
Stage organization also prevents the most common workflow failure: skipping steps. When prompts live in a random folder of saved chats, it’s easy to jump straight from an idea to a draft and skip research, intent analysis and the brief. When they’re arranged in sequence, the gaps are visible.
What did the great production systems figure out?
Three historical systems show why staged workflows produce consistent output.
Ford’s moving assembly line. In 1913, Ford introduced a moving assembly line for the Model T at its Highland Park plant. Breaking production into defined stations dramatically cut the time needed to assemble a car.
The McDonald brothers’ Speedee Service System. In 1948, Richard and Maurice McDonald redesigned their San Bernardino restaurant around a short menu and a kitchen organized into specialized stations. Speed and consistency came from the workflow, not from any one cook.
The Toyota Production System. Developed in postwar Japan, it paired staged production with the expectation that any worker could stop the line to fix a problem. Quality was built into each stage instead of inspected at the end.
Content isn’t cars or burgers. But the principle travels: defined stages, clear handoffs and quality checks built in along the way.
What are the ten stages?
Here’s the full cycle: what each stage does, what it hands off and how many prompts the Prompt Pack devotes to it.
Stage 1: Keyword and topic research (15 prompts)
Job: Generate and shape topic candidates around a business, audience and offer.
Hands off: A shortlist of topics with working queries and rough angles.
Your call: Choose topics that fit the business and that you can cover with real authority. A model can propose. Only you know what you can credibly say.
Stage 2: Search intent and AI-answer analysis (12 prompts)
Job: Work out what a searcher actually wants and how AI answers tend to frame the topic today.
Hands off: An intent statement and notes on the angle your piece should take.
Your call: Check intent against what customers actually ask you. Real questions beat inferred ones.
Stage 3: Content briefs and outlines (15 prompts)
Job: Turn a topic and intent into a structured brief: audience, purpose, key questions, outline and required facts.
Hands off: A brief a writer, human or model, can draft from.
Your call: Approve the brief. It’s the cheapest point in the workflow to change direction.
Stage 4: Drafting and brand voice (15 prompts)
Job: Produce a first draft from the brief, in the brand’s voice, using supplied facts.
Hands off: A complete draft with unsupported claims flagged.
Your call: Supply facts through the bracketed variables. Read the result as an editor, not a fan.
Stage 5: On-page SEO (15 prompts)
Job: Titles, meta descriptions, headings, internal link suggestions and other on-page elements.
Hands off: A draft with its on-page package attached.
Your call: Choose among options, and make sure the title promises only what the page delivers.
Stage 6: AEO, FAQ and answer-ready snippets (15 prompts)
Job: Write concise, direct answers to specific questions, built to be quotable.
Hands off: Answer blocks and FAQ entries for the page or for dedicated answer pages.
Your call: Make every answer specific to the business, not generic.
Stage 7: Content refresh and pruning (10 prompts)
Job: Review existing content for staleness, gaps and overlap, and plan updates.
Hands off: Refresh notes and keep, merge or retire decisions.
Your call: Make the final decision. Models are good at spotting staleness. Business priorities are yours.
Stage 8: Schema and technical helpers (8 prompts)
Job: Draft structured data and help with technical elements tied to the content.
Hands off: Draft JSON-LD and notes for whoever manages the site.
Your call: Validate schema with a testing tool before deployment, and make sure it matches the visible page.
Stage 9: Repurposing for email, Pinterest and social (10 prompts)
Job: Adapt published content for other channels without copying it wholesale.
Hands off: Channel-specific versions ready to schedule.
Your call: Make sure each version suits the channel and your real presence there.
Stage 10: QA and fact-check (5 prompts)
Job: Review drafts for unsupported claims, inconsistencies and invented details.
Hands off: A list of items to source, soften or strike, and a cleaner draft.
Your call: Decide every flagged item. This is the gate.
Why does the QA stage come last but matter most?
Because it’s the last chance to catch a confident error before a reader does. The Prompt Pack’s own quick start is direct about this: run a Category 10 prompt on every draft before publishing. That one habit is the clearest difference between AI-assisted content that builds trust and content that quietly spends it.
Notice that QA has the fewest prompts. It doesn’t need many. It needs to run every time.
Do you have to use all ten stages for every piece?
No. A short FAQ answer might go from stage 2 to stage 6 to stage 10. A refresh starts at stage 7. A pillar article might touch every station. The sequence isn’t there to force every piece through every stage. It’s there so that when you skip one, you’re doing it on purpose.
One rule worth keeping: stages 3 and 10 are never optional. Every piece gets a brief, however short, and every piece gets a QA pass.
Where do teams usually break the cycle?
Six familiar failure points:
- Skipping intent analysis, then writing a great answer to the wrong question.
- Treating the outline as the brief. An outline says what goes where. A brief says why, for whom and with which facts.
- Drafting without variables, so the model invents the specifics.
- Letting on-page elements overpromise. A title that claims more than the page delivers costs trust.
- Repurposing by copy-paste, which reads as lazy on every channel at once.
- Skipping QA under deadline, exactly when it matters most.
How do you set the library up for a team?
The Prompt Pack ships as a PDF, a Notion-ready CSV (ID, Category, Title, Prompt and Tags) and a Markdown version. For teams, the CSV is the workhorse. Import it into Notion as a database, add your own columns for status, owner or client, and filter by category. The library becomes a stage-by-stage menu.
For solo practitioners, the Markdown version drops neatly into Obsidian or any notes app, where prompts can sit beside your fact sheets and briefs. The PDF adds a short prompting primer and a worked walkthrough on a fictional business that shows filled variables, sample output and the fact-check edit, so you can see the whole line running before you put real work on it.
How does the cycle change a typical week?
Here’s what a week can look like for a freelance content marketer handling two clients:
- Monday: stages 1 and 2 for both clients’ next topics. Shortlists and intent notes go to the clients for a quick yes.
- Tuesday: stage 3 briefs, approved by end of day.
- Wednesday and Thursday: stages 4 through 6, drafting with each client’s fact sheet feeding the variables.
- Friday morning: stage 10 on every draft. Flagged items get sourced, softened or struck.
- Friday afternoon: stage 9 adaptations for last week’s published piece, and stage 7 notes on one older page per client.
Nothing about that week is heroic. That’s the point. A staged line turns content work from a series of creative emergencies into a schedule you can keep, delegate and price.
Quick answers
Is a staged workflow slower than just asking for a draft?
For a single piece, sometimes a little. Across a month of content, it’s faster, because you spend far less time fixing drafts built on guesses.
Can different people run different stages?
Yes, and that’s a strength. One person researches and briefs, another drafts, a third runs QA. The handoffs are defined, so nothing gets lost between desks.
Does this work with any AI assistant?
The prompts are tool-agnostic and work with the major chat assistants. The stages work with any tool, including a pencil.
Build the line once
A content cycle that depends on improvisation produces improvised results. A cycle with defined stations, clear handoffs and a firm QA gate produces work you can schedule, delegate and stand behind.
Get the SEO & Content Prompt Pack: 120 tool-agnostic prompts across ten stages, from first research to final fact-check, in PDF, Notion CSV and Markdown, for $12. Set up the line once, and every piece after that starts at the right station.
You’ll find the ten-stage library on Content & SEO Workflows.
Leave a Reply