How to create a narrative visual with Copilot in Power BI

João Barros 06 de July de 2026 4 min read

A narrative visual with Copilot in Power BI turns the numbers in a report into clear, natural-language text. Instead of interpreting several charts on their own, the reader gets a summary that points out highlights, drops and trends. It is a fast way to make reports more accessible to people who are not used to reading data, and to save time for whoever prepares presentations.

Prerequisites

An up-to-date Power BI Desktop or the Power BI Service (web version).
A paid Fabric capacity (or Power BI Premium) with Copilot enabled by your administrator.
A report with a few visuals already built on the data you want to summarise.
Edit permission on the report (or build permission on the semantic model).

Step 1: Prepare the page with visuals

The narrative visual reads what is on the report canvas, not the semantic model directly. So the clearer your visuals are, the better the summary will be. Open (or create) a page with the charts and tables you want to explain — for example, sales by month, by region and by category. Give each visual and its axes descriptive names, because Copilot uses those titles to understand the context and produce correct sentences.

Tip: a chart named "Sales by month" produces a far better summary than a chart with the generic title "Chart 1".

Step 2: Add the narrative visual

With the page open, go to the Visualizations pane on the right and select the Narrative icon. In the Choose a narrative type window, pick the Copilot option to use the AI-generated summary instead of the classic smart narrative. Copilot analyses the visuals on the page and proposes a first draft automatically.

Step 3: Choose what Copilot will summarise

In the Create a narrative with Copilot dialog, set the scope of the summary: the whole page or only some specific visuals. Select the option you want and click Create. Within a few seconds a paragraph appears with the main numbers and changes, already formatted inside the visual.

Step 4: Adjust the text with prompts

The first result is only a starting point. In the Adjust your summary with Copilot area you can rewrite the text with natural-language instructions or use the suggested prompts. The trick is to give concrete requests, for example:

Summarise total sales and highlight the best-performing month.
Explain the change versus the previous period as a percentage.
Use a more formal tone and limit the summary to three sentences.

After each instruction, Copilot rewrites the text. Keep refining request by request until the summary says exactly what you need — no more, no less.

Step 5: Format and check the sources

As with any visual, you can adjust the font, colour and size in the Format pane. Notice the numbered footnotes inside the text: when you click one, Power BI highlights the visual that statement came from. This is very useful to confirm that every sentence in the summary has a real origin in the data.

Check the result

To confirm everything is right, filter the page — for example, select a year or a region — and watch whether the summary text updates on its own. The narrative visual is dynamic and follows the page filters and selections. Also check that the values mentioned match the charts and that the footnotes point to the correct visuals. If a number looks odd, review the name and filters of the matching visual.

Conclusion

With just a few clicks you now have a text summary that updates with the data and makes the report far easier to read. The natural next step is to try several prompts for the same report and compare which one produces the clearest explanation for your audience. Which report on your team would benefit first from an automatic summary like this?