How to write effective prompts for Copilot in Power BI

João Barros 07 de July de 2026 4 min read

A good prompt is the difference between a generic answer and a result that saves you minutes of work. Learning to write effective prompts for Copilot in Power BI helps you generate more accurate summaries, measures, and report pages on the first try, without rephrasing your request five times. Writing your requests well is now as useful a skill as knowing how to build a measure or design a visual. Below is a simple, repeatable method, with example prompts you can copy and adapt to your own data.

Prerequisites

A report or semantic model in Power BI (Desktop or Service) with the Copilot pane available.
A tidy data model, with clearly named tables, columns, and measures.
A concrete idea of what you want: a summary, a measure, a report page, or an answer to a question.

Step 1: Start with the goal and the format

The golden rule is to ask for one thing at a time and state the format you expect. A vague prompt almost always returns a vague answer, because Copilot has to guess what you meant. Say what you need, for which period, and in what format — bullets, a table, or a number of sentences. Notice the difference between the two requests:

Less effective: "Tell me about sales."
More effective: "Summarise total sales by region for the last
quarter in 3 bullet points, from highest to lowest."

Step 2: Give context about your data

Copilot in Power BI knows the structure of your semantic model, but it doesn't know your business intent or which measure is the "right" one. By naming the actual tables, columns, and measures, you steer the answer and reduce mistakes. Be specific about the calculation and the formatting you want:

Create a DAX measure that calculates gross margin from
[Revenue] and [Cost] in the Sales table. Format it as a
percentage with one decimal place.

Step 3: Ask one step at a time and iterate

Large requests tend to produce confusing answers. Break the task into small steps and refine the previous answer instead of starting over. This layered conversation gives you far more control over the final result:

1) "Create a report page about sales performance."
2) "Add a card with total revenue at the top of the page."
3) "Show only the top 5 products by revenue in the bar chart."

Step 4: Define the audience and tone for summaries

When you ask for a summary, say who it's for and how much detail you want. The same report can produce a one-line summary for executives or a detailed explanation for the analytics team. Naming the audience avoids answers that are too technical or too shallow:

Summarise this report for executives in 4 sentences.
Highlight the changes versus last month and avoid technical jargon.

Tip: save the prompts that work well in a notes file. Over time, build a small library of reusable prompts for the tasks you repeat every week.

Step 5: Always review the answer

Copilot speeds up your work, but the responsibility is still yours. Check the numbers, confirm that the correct measures and columns were used, and validate the generated DAX logic before you publish it. If something looks off, adjust the prompt with more context and generate again.

Check the result

You recognise a good prompt by its result: the answer does exactly what you asked, uses the right fields, and barely needs corrections. A practical test is to count how many times you have to rephrase. If you need three or four attempts for the same thing, the original prompt was too vague — add goal, format, and context, and try again.

Conclusion

With a clear goal, context about your data, and step-by-step iteration, you turn Copilot in Power BI into a reliable assistant rather than a generator of generic answers. The next step is to start your own prompt library, one for each task you repeat day to day. Which prompt will be the first you save?