The Five Things a Useful Request Usually Needs

Turn a vague wish into directions that someone else can follow.

Imagine asking a neighbor, “Can you help with my event?”

The neighbor might offer to move tables. You might need someone to write an invitation. Both interpretations make sense. Neither person has enough information to know whether they agree.

The same problem appears in an AI conversation. “Help with my event” could produce a budget, a checklist, a speech, or twelve suggestions involving a balloon arch. The answer might look thoughtful while missing the job you actually need done.

A useful request closes that gap. It tells the assistant what you want, supplies the information it needs, and gives you a way to check the result.

You do not need special vocabulary. Five ordinary questions will usually get you started.

1. What should exist when this is finished?

Begin with the outcome. Name something you can inspect: an invitation, a comparison table, a study quiz, a corrected paragraph, or a list of questions to ask a supplier.

“Help me organize” describes an intention. “Turn these notes into a checklist for opening our booth” describes an output.

Consider a fictional community book swap. Elena wants an invitation, but she also has unanswered questions about parking and refreshments. If she asks for “an event plan,” the assistant may spend its effort on those questions.

Her first sentence should be narrower:

Draft an invitation for our community book swap.

That sentence gives the work a destination. It also keeps a small writing task from growing into an entire project.

If you cannot name the output yet, discovery can be the output: “Help me identify which three decisions I need to make before planning this event.”

2. What background changes the answer?

Context means relevant background. It includes the audience, situation, purpose, and anything the assistant might otherwise misunderstand.

For Elena, the invitation is going to neighbors who have never attended a book swap. They need to understand how it works. A message to experienced volunteers would need different information.

She adds:

Write for neighbors who have never attended. Explain the exchange in plain language and make the invitation welcoming.

Useful context changes a decision about the answer. Her favorite color probably does not matter. The fact that attendance is free does.

One way to choose background is to ask, “Would a competent person do this differently if they knew this fact?” If yes, include it. If no, leave it out unless the task calls for it.

3. What information should the assistant use?

Inputs are the material the assistant will work on. For an invitation, they are event facts. For a summary, they are the source text. For editing, they include your draft.

Elena supplies this fictional fact list:

She labels the list “confirmed event facts.” She does not mix it with an old brainstorming note about selling tickets.

The distinction matters. An assistant can turn both facts and suggestions into smooth prose. Your job is to show which is which.

If the address is not available, say so. “Street address not yet supplied” is more useful than leaving the assistant to decide whether it should invent one.

4. What limits must the result respect?

Constraints describe practical boundaries: length, time, budget, format, allowed sources, or actions that belong outside this task.

For Elena:

Keep the invitation under 120 words. Use only the confirmed facts. Do not add refreshments, parking instructions, an address, or a registration deadline.

Those exclusions address likely mistakes in this particular task. There is no need for a page of rules about every imaginable error.

Positive directions help too. “Use short paragraphs” gives the assistant something to do. “Do not be confusing” leaves it guessing.

When two limits conflict, decide which matters more. If essential instructions will not fit in a tiny word limit, ask the assistant to flag the conflict. Do not quietly sacrifice meaning to satisfy a number.

5. How will you know it worked?

Success criteria are observable checks. They turn “make it great” into a decision you can make.

Elena chooses four:

• The date, time, and place match the notes.
• Readers understand both five-book limits.
• Readers understand that bringing books is optional.
• The invitation ends with the requested reply action.

A lively sentence does not compensate for changing “up to five” into “at least five.” The checks protect what matters.

Google's prompting guidance supports clear instructions and treating prompts as something to refine through observation. The five-question brief here is a practical teaching template for applying that general idea. Google: Prompt design strategies.

Put the pieces together

Here is Elena's complete request:

Draft an invitation for our community book swap for neighbors who have never attended. Explain the exchange in plain, welcoming language.

Confirmed facts: October 24, 2026, 10 a.m.–noon, Maple Hall, free entry. People may bring up to five books and take up to five books. Bringing books is optional. Ask interested readers to reply.

Keep it under 120 words. Use only these facts; do not add an address, refreshments, parking details, or a deadline. Success means every event detail is correct, both book limits are clear, and the optional contribution is explicit.

That is enough to begin. A possible draft would say:

Join us for a free community book swap at Maple Hall on October 24, from 10 a.m. to noon. Bring up to five books to share and take up to five books home. You are welcome even if you have no books to bring. Interested? Reply to this invitation.

Elena can now check the output against the brief. She still needs to supply the street address before using the invitation where readers would need it.

Students: define the learning you want

A student might write, “Help me with fractions.” A clearer learning request would be:

Help me practice adding fractions with different denominators. I can already simplify fractions. Give me one practice problem at a time, wait for my attempt, and explain my first mistake. Use new problems, not my graded assignment. Success means I can solve three new examples and explain the steps myself.

The goal is now a skill the student can demonstrate. If class rules limit AI assistance, those rules belong in the constraints. The assistant cannot supply permission on a teacher's behalf.

Your exercise

Rewrite these three requests using the five questions:

1. “Make this email better.”
2. “Help me study.”
3. “Plan our project.”

Choose one. Try the vague version and your revised version in separate conversations with the same underlying material. Compare the outputs against your written checks. One comparison gives local feedback; it does not prove that a phrasing will always perform better.

Then shorten your improved request. Remove words that do no work. Keep the information another person would need to understand and check the job.

Completion check: Someone who has not watched your conversation can explain what you requested and tell whether the result meets it.

Previous article: 010 — Your First Week with AI.