September 21, 2026
Ask a vague question, get a vague answer. That is not a flaw in the tool; it is a gap in the request. Almost every disappointing result traces back to one of a handful of habits, and each one takes about thirty seconds to fix. Below are the ten mistakes beginners make most often, each with a before-and-after so you can see the difference right away. Read it once, then keep it nearby the next time an answer comes back slightly wrong.
1. Asking for everything at once
Mistake 1: Three questions in a single prompt
"Summarise this report, turn it into a social post, and suggest five headlines." That sentence asks for three deliverables, in three formats, at three different lengths. Something has to give, and it is usually the part you cared about most. The model spreads its attention thin and returns three mediocre results instead of one good one.
The fix is one task per prompt. If you genuinely need all three, ask for a plan first, then work through it step by step.
Before: "Summarise this report, write a post about it, and give me headlines."
After: "Summarise this report in five bullets, each under 15 words. I will ask for the post separately."
Mistake 2: The vague one-liner
"Write about pricing" tells the tool nothing about who is reading, what you want them to do, or how deep to go. The default is a broad overview written for nobody in particular, which is exactly what you get.
Add four things: audience, purpose, angle and one boundary.
Before: "Write about pricing."
After: "Write 300 words explaining tiered pricing to first-time freelance designers who have never priced a project. Focus on what to charge for revisions. No formulas, no psychology."
2. Withholding context
Mistake 3: No role, no reader, no situation
Setting a role is not theatre. Telling the tool it is editing for a sceptical finance director changes the vocabulary, the structure and what gets left out. So does naming the reader and what they already know.
Before: "Explain our new process."
After: "You are briefing a new team member on day one. Explain our approval process in plain language. They know the software already but not who signs off on what. Skip the history."
Mistake 4: Describing the style instead of showing it
"Friendly and conversational" means something different to everyone. A sample does more work than a paragraph of adjectives. Paste 40 to 60 words of writing you like and ask for that rhythm and level of formality.
If you have no sample, describe it with rules rather than moods: sentences under 20 words, contractions fine, no exclamation marks, no rhetorical questions.
3. Saying nothing about the output
Mistake 5: Skipping the format
If the result is going into a slide, ask for a table. If it is going into an email, ask for a subject line and two short paragraphs. Otherwise you get a wall of text you have to reformat by hand, which defeats the point.
Worth naming every time:
- Structure: prose, bullets, numbered steps, a table, a dialogue.
- Labels: whether you want headings, and how short they should be.
- Length: a word count or a number of items, not "short".
- Tone: plain, formal, dry, warm — pick one.
- What to include: examples, a specific point of view, a counterargument.
- What to leave out: no intro, no summary, no caveats.
Mistake 6: Ignoring length and scope
"Brief" is not a measurement. Say "150 words" or "five bullets, one line each". Scope matters just as much: "cover the three most common objections and stop" is a real instruction, and it prevents a sprawling answer that answers questions you never asked.
4. Instructions that cannot be measured
Mistake 7: Asking for "good"
"Make it good" is a wish, not a brief. Swap adjectives for rules. Instead of "make it engaging", write "open with a one-sentence scene and keep paragraphs under three lines". Two or three concrete rules will beat a page of praise-words every time.
Mistake 8: Constraints that cancel each other out
"Be thorough but keep it brief." "Professional but playful." "Cover all ten points in 100 words." When instructions conflict, the tool picks one and you get something half-right. Decide which one matters more and say so: "Prioritise brevity. If you must cut, cut the examples."
5. Treating the first draft as final
Mistake 9: Starting over instead of steering
Rewriting the whole prompt throws away everything that worked. Name the problem and keep the rest.
Instead of: "That's not right, try again."
Try: "Keep the structure. Rewrite paragraph two — it is too abstract. Give me one specific example instead."
When you are stuck on a section, ask for options rather than a verdict: "Three different opening lines, same tone." Each round should be a small edit, not a fresh start.
Mistake 10: Dragging a long, messy conversation
After twenty exchanges, old instructions still linger and quietly contradict the new ones. When an answer drifts, start a fresh prompt with a compact brief and carry over only the parts that worked.
Goal: the one thing you want produced.
Reader: who it is for and what they already know.
Context: the details only you have — a sample, a constraint, a real deadline.
Format and length: structure plus a number.
Rules: two or three measurable ones, plus what to leave out.
Save the briefs that land. A folder of eight working prompts will save you more time than any clever trick you read online.
The sixty-second pass before you hit send
Run every prompt through this list first. It takes a minute and prevents most of the ten mistakes above.
- State the one task in a single sentence.
- Name the reader and what they already know.
- Add the context only you can provide.
- Specify the format and a number for the length.
- Replace every quality adjective with a rule.
- Read it as a stranger. Would you know exactly what to produce?
None of this is complicated. It is the difference between handing someone a task and handing them a guess. Write the brief you would want if you were the one doing the work, keep the versions that land, and the next prompt will be easier than the last.
Photo: MART PRODUCTION / Pexels
