A checklist is only useful if every item earns its place, so this one comes with a one-line justification per check. Use it as a working tool: run a prompt against it before you ship the output, or build it into your drafting habit. The items are grouped by stage—define, build, verify, maintain—so you can apply them in the order the work actually happens.
This is not a substitute for understanding the underlying practice. If the reasoning behind any item is unclear, the linked companion articles explain it in depth. The checklist is meant to operationalize that understanding so you do not have to hold all of it in your head every time. Think of it the way a pilot thinks of a preflight list: not because the pilot has forgotten how to fly, but because under pressure even experts skip steps, and a list catches the skip before it costs anything.
Print it, paste it next to your editor, or memorize the four groups. The goal is to make audience-adaptive design a reflex rather than an effort you have to summon. The list is deliberately ordered to follow the flow of real work: you define the reader, build the prompt around them, verify the output serves them, and maintain what works for next time. Run it top to bottom and the sequence does the thinking for you.
One caution before the items. A checklist tempts you to treat each box as a yes-or-no formality. Resist that. The value is in actually doing what each item asks, not in marking it complete. A box checked without thought is worse than no box, because it creates false confidence that the work was done.
Before You Write: Define the Reader
These checks happen before a single instruction is drafted. Skipping them undermines everything after.
The definition checks
- Have you written the audience down? Holding it in your head lets it stay vague; typing it forces specificity.
- Did you name the reader's expertise level? It drives nearly every later choice about depth and vocabulary.
- Did you state the reader's immediate goal? What they want to do shapes what the answer should foreground.
- Did you note their jargon tolerance? This single attribute decides whether terms get used, defined, or avoided.
- Did you describe what they specifically lack, not a stereotype? "Beginner" is a label; "new to this domain but experienced elsewhere" is a target. The difference is covered in Mistakes That Quietly Erode Prompt Reliability.
While You Build: Set the Dials
These checks govern the prompt's construction. They turn the reader profile into concrete instructions.
The construction checks
- Does the audience appear before the task? Leading with the reader anchors every later instruction to them.
- Did you set vocabulary explicitly? "Use plain language" or "use industry terms freely" is a dial; "be clear" is not.
- Did you set depth and entry point? State how deep to go and where to begin rather than letting the model default.
- Did you decide what to include and omit? Real adaptation changes substance, not just tone.
- Did you attach a calibration sample? One example in the target voice steers register better than abstract description. The full lever set is in Writing One Prompt That Speaks to Many Readers.
Before You Ship: Verify the Fit
These checks happen on the output. They catch the failures that look fine until a real reader hits them.
The verification checks
- Did the prompt include a self-check for fit? A built-in instruction to confirm register catches drift before you read.
- Did you read the output as the actual reader? Reading as the author hides every adaptation gap.
- Did simplification preserve accuracy? Check that the easier version is still true, not just easier. This applies the verification mindset from Models Are Learning to Catch Their Own Mistakes.
- Did you check for register drift in the later paragraphs? Models often revert to default voice as text accumulates.
- Would the reader feel respected, not patronized? Condescension is a silent failure the author rarely notices.
After It Works: Maintain the Asset
These checks keep a good prompt good and prevent it from being misused later.
The maintenance checks
- Did you save the prompt with its audience profile? A working prompt is an asset; document what it was tuned for.
- Did you note where it stops working? A prompt for novices fails for experts; record the boundaries.
- Did you test it at the edges of its audience range? A prompt tuned for the median can fail at both extremes.
- Did you decide whether one prompt suffices or you need to branch? Wide audiences sometimes require separate prompts.
How to Use This Without Slowing Down
A checklist that doubles your effort gets abandoned. Here is how to keep it light.
Match the rigor to the stakes
For low-stakes drafts, run only the define and verify groups. For anything consequential, run all four. The checklist scales with how much a misfit would cost.
Internalize the groups, not the items
Over time, you stop reading the list and start thinking in its four stages—define, build, verify, maintain. The items become reflexes. The sequence that builds this habit is laid out in The Sequence That Turns a Vague Audience Into a Working Prompt.
Run the failing items, not all of them
After a few weeks you will know which checks you reliably pass and which you keep tripping on. Spend your attention on the ones you fail. If you never skip the audience definition but you constantly forget to read as the reader, prune the list down to your own weak spots. A personalized checklist gets used; an exhaustive one gets ignored.
Adapting the Checklist to Your Context
No single list fits every kind of content. Treat this one as a starting template you tailor.
Add domain-specific checks
If your content has its own failure modes—regulatory language that must stay precise, accessibility requirements, brand voice constraints—add items for them. The four groups give you a place to slot new checks: a brand-voice check belongs in build, a compliance check in verify.
Drop checks that never fire
If a check has never caught a problem in your work over many uses, it may not apply to your context. A list that carries dead weight loses credibility, and a checklist you do not trust is one you stop running. Keep it lean enough to stay honest.
Frequently Asked Questions
Do I have to run every item every time?
No. Match the rigor to the stakes. For quick, low-risk output, the define and verify groups are enough. For anything consequential, run all four. Forcing the full list on trivial tasks is how checklists get abandoned.
Which item catches the most problems?
Reading the output as the actual reader rather than the author. It surfaces condescension, assumed knowledge, and wrong-level depth that no syntactic check finds, because the author always reads with context the reader lacks.
Why is "describe what they lack, not a stereotype" on the list?
Because labels like "beginner" carry baggage that produces patronizing output. Describing the specific gap—new to this subject, expert in adjacent ones—gives the model a respectful, precise target and avoids the condescension that crude labels invite.
How do the maintenance checks help if a prompt already works?
They prevent a working prompt from being reused where it does not fit. A prompt tuned for novices fails for experts, so recording its audience and boundaries stops future misuse and saves you from rediscovering its limits the hard way.
Can I turn this checklist into a template?
Yes. The define group maps cleanly onto a fillable audience profile, and the build group onto a prompt scaffold. Encoding the checklist as a reusable template is the natural next step once the items feel routine.
Key Takeaways
- Define the reader before writing: expertise, goal, jargon tolerance, and the specific gap rather than a stereotype.
- Build with explicit dials—audience first, vocabulary, depth, entry point, substance decisions, and a calibration sample.
- Verify on the output: self-check for fit, read as the reader, confirm simplification kept accuracy, and watch for drift.
- Maintain working prompts by saving them with their audience profile and documented boundaries.
- Scale the checklist to the stakes and internalize the four groups so they become reflexes rather than effort.