Topic 2: Prompting 101: Bad prompts
Prompting 101: Bad Prompts
A lot of disappointing AI output starts with a simple problem: the AI wasn’t given enough direction.
When we write a vague prompt, AI has to guess what we mean. It may guess the audience, the purpose, the level of detail or even the problem we are trying to solve.
For Product Managers and Business Analysts, this matters because our work is usually full of context. A user story, product strategy or requirements analysis cannot be useful if the AI does not understand the product, users or situation behind it.
What makes a Prompt Bad?
A bad prompt is not necessarily grammatically bad. It is usually just too vague or too generic for the task.
Consider:
“Write a product strategy.”
There are many unanswered questions.
What product? What market? What stage is the product at? Who are the customers? What business problem are we solving? What does the strategy need to cover?
Without this information, AI has to fill in the gaps itself.
The result may look impressive, but it is likely to be generic.
Watch out for vague instructions
Some prompts look perfectly reasonable until you think about what the AI actually knows.
For example:
“Give me user stories.”
For what product? Which feature? Which users? What problem are they trying to solve?
Or:
“Make this better.”
Better in what way? More concise? More persuasive? More professional? Easier for a senior stakeholder to understand?
Or:
“Summarise this.”
Who is the summary for? What information matters most? How long should it be?
The lesson is simple. Do not make AI guess what you mean when you can tell it.
Build better prompts step by step
You do not need to write a huge prompt. Start by answering a few useful questions.
1. What do you want?
Define the task or output clearly.
Instead of simply saying “Help with this research”, tell AI whether you want it to analyse, summarise, compare, identify themes or recommend next steps.
2. Who is it for?
Think about the audience.
A summary for a senior executive may need to be very different from something written for a product team or development team.
3. What style or format do you need?
Tell AI how you want the response presented.
You might ask for:
- three recommendations
- a 200-word summary
- a comparison table
- user stories
- a checklist
- a simple explanation
This makes the output much easier to use.
4. What context is important?
Give AI the background it needs.
For PM and BA work, this might include the product, users, business objective, research, requirements, data or constraints.
You do not need to include everything. Include the information that helps AI understand the task properly.
Why better prompts produce better results
Better prompting is not about making prompts longer.
It is about making them clearer, more specific and better structured.
When AI understands your purpose, audience, context and expected output, it has a much better chance of giving you something relevant and useful.
DSM Tip
Great prompts lead to great results.
Before sending a prompt, ask yourself: Have I given AI enough information to understand what I actually want?
If the answer is no, add the missing context before you hit send.
