Topic 3: How to write effective prompts

How to write effective promts

How to write effective prompts

Writing an effective prompt does not mean learning complicated formulas or finding the perfect words.

For a Product Manager or Business Analyst, it is much simpler. You need to help the AI understand what you want it to do, what it needs to know and what a useful result should look like.

A simple structure you can remember is:

Task + Context + Requirements + Output

You can then add examples and continue refining the response when the task needs more guidance.

1. Define the task

Start by telling the AI exactly what you want it to do.

Use clear action words such as:

  • analyse
  • summarise
  • compare
  • create
  • identify
  • review
  • prioritise
  • improve

Instead of:

“Here are some customer interviews. What do you think?”

Try:

“Analyse these customer interviews and identify the five most common customer problems.”

The second prompt gives the AI a much clearer job.

2. Give relevant context

Next, give the AI the information it needs to understand your situation.

For PM and BA work, this might include:

  • what the product or service is
  • who the users are
  • the business problem
  • the objective you are trying to achieve
  • relevant requirements
  • customer research
  • previous decisions
  • important project information

You do not need to give AI everything you know. Give it the context that matters for the task.

3. Add requirements & boundaries

Now explain anything the AI needs to consider while completing the task.

For example:

“Only use information from the customer research provided.”

Or:

“Do not suggest technical solutions yet. Focus only on identifying the customer problems.”

You can also specify assumptions, rules, constraints, criteria or things the AI should avoid.

This helps keep the response from becoming too broad or moving in a direction you didn’t want.

4. Define the output

Tell the AI what you want the answer to look like.

For example, you could ask for:

  • a table
  • a summary
  • user stories
  • acceptance criteria
  • a list of questions
  • recommendations
  • options with pros and cons
  • risks and assumptions

If you are going to use the output in a workshop, document or backlog, asking for the right structure can save you a lot of additional work.

5. Add examples when helpful

Sometimes it is easier to show the AI what good looks like.

If you already have a well-written user story, requirement, product description or another example, include it and ask the AI to follow a similar structure or level of detail.

Examples are particularly helpful when you want a specific format or style.

6. Review the response

A good prompt does not guarantee a perfect answer.

Read the response and ask yourself:

  • Has the AI understood the task?
  • Has it made assumptions?
  • Is anything important missing?
  • Are the facts correct?
  • Is the reasoning sensible?
  • Is the output actually useful?

Your professional judgement is still important.

7. Refine and continue

You don’t need to start again every time the response isn’t quite right.

Continue the conversation.

You might say:

“You have assumed that these users are existing customers. They are actually new customers. Review your analysis with this information.”

Or:

“Good start. Now identify the assumptions behind each recommendation.”

Each follow-up gives the AI more direction and helps you improve the result progressively.

Putting it together

A practical prompt might look like this:

  • Task: Analyse the customer interview notes and identify the main customer problems.
  • Context: We are redesigning the onboarding journey for new users of our B2B product.
  • Requirements: Only use evidence from the interviews. Group similar problems together and highlight anything that needs further research.
  • Output: Give me a table with Customer Problem, Evidence, Frequency and Questions to Investigate.

Notice that the prompt isn’t particularly complicated.

It simply gives the AI enough direction to understand the job.

DSM Tip

Task + Context + Requirements + Output is a strong starting point for most prompts.

Don’t worry about creating the perfect prompt in one attempt. Start clearly, review what comes back and continue improving the result through conversation.

Shruti DSM