Build an interactive data story in minutes

Start with one sentence that explains what your audience should take away. This gives the Connector a clear direction and helps shape every decision that follows, from chart type and time range to color and annotation.

You can also ask the Connector to sharpen your brief and check whether the dataset supports the story you want to tell.

I want to show how G7 nations industrialized at very different rates and have since reduced emissions unevenly. 
My audience is a general reader interested in climate progress.

Write a brief for a three-slide data story covering the industrial rise, the post-1990 policy era, and recent divergence. My audience is a general reader interested in climate change progress.

* Confirm understanding of the brief and wait for next steps.

Ask the LLM to read your dataset and summarize what it finds before building anything. This helps catch issues early, such as unclear columns, unexpected date ranges, inconsistent entity names or values that could be misread later in Flourish.

Audit this dataset and report back: (1) column names and types, (2) which entities are present and how many rows each has, (3) earliest and latest year per entity, (4) value range for the main metric, (5) any nulls, gaps, or anomalies worth flagging. 

* Summarize your findings and wait for my confirmation before moving on. 

Depending on your story, it can be useful to narrow the dataset before creating charts in Flourish. Here, the data goes back to 1750, but the industrialization story is clearer from around 1800.

Ask the Connector to filter the rows first, so the final charts stay focused without needing manual spreadsheet edits.

Filter this dataset to only include rows where Year is 1800 or later. Keep all columns: Entity, Code, Year, CO₂ emissions per capita (tonnes).

* Flag when the filter is complete and wait for next steps.

Before building anything, ask the LLM which visualizations would create the strongest narrative arc from your dataset. A three-chart story might include historical context, a modern trend and a present-day comparison.

The Connector can suggest Flourish chart types, time ranges and story roles for each chart, helping you avoid building the wrong option first.

Based on the filtered dataset, what Flourish chart types would best show the compelling insights? Suggest 2-3 charts per angle, explain why, add a confidence level (1-10), fact-check claims, and note what each slide should make the reader feel or understand.

* Make a set recommendation and wait for next steps

Once you’ve chosen the story angles, use the Flourish Connector to build the set of charts that best support the narrative.

A key benefit is that the Connector can handle the data setup for each chart. It reads the structure needed for the selected Flourish template, then reshapes your data to match — including column reordering, renaming and binding. What would usually take extra time in a spreadsheet or code can happen as part of the same prompt you use to create the chart.

Create all the charts with highest confidence for each angle in sequence.

Check each chart before assembling the story. Does it support the point you want to make? Does the chart type feel right? Is anything missing, unclear or too detailed?

If a chart doesn’t work, ask your LLM to revise it or suggest an alternative. The Flourish Connector can update the existing chart or build a replacement, so you can improve the set before moving into final polish.

For the uneven progress chart, create a slope chart to compare each country's peak and the latest year. Use this new chart as part of the data story. 

* Confirm the build and update the list of charts in the data story

Use line and range highlights, point markers, or reference lines to draw attention to the key point in each chart. Ask the LLM to draft the copy, then use the Connector to apply the updates in the right format for each visualization.

This is where the Connector is especially useful: instead of editing each chart manually, you can apply a full set of annotation changes in one prompt.

For the Industrialization Gap small-multiples grid, add a line highlight to each panel marking the peak emissions year in the underlying dataset for that country. Use a consistent dark gray label for all highlights, format the date as YYYY, and set the label max width to 20.

For the Staggered Peaks combined line chart, add three line highlights: Kyoto Protocol at 1997 in dark gray, Paris Agreement at 2015 in blue, and COVID-19 at 2020 in red. Set each label color to match its corresponding annotation color, format all dates as YYYY, and set the label max width to 30.

Color is one of the fundamentals of clear data visualization. For it to work across a story, it needs to be consistent: the same country should use the same color in every chart.

Use the Connector to apply the same pinned color overrides across all three charts, then add chart-specific emphasis where the story needs it.

Apply these pinned color overrides to all three charts: Canada → #fc8d59, France → #abd9e9, Germany → #74add1, Italy → #c2e0f4, Japan → #fee090, United Kingdom → #4575b4, United States → #d73027. 

Then in the Staggered Peaks combined line chart, highlight United Kingdom and United States — they represent the two extremes. 

In the Uneven Progress slope chart, confirm that line and label colors match the pinned overrides and correct any that don't.

If color overrides don’t apply as expected, the chart may need a different data shape. For example, Chart 1 uses a long-format dataset, where all countries sit in a single country column.

Ask the Connector to check the template requirements, reshape the data if needed, and reapply the pinned colors.

For the color overrides to work in Chart 1, convert the dataset to a wide format so each country is an individual series with it's own color.

Before adding the charts to the Story editor, review the full set for consistency. Make sure titles, source credits, annotations and visual highlights follow the same structure, so the story feels intentional from start to finish.

Update the charts with these:

1. Add the same source credit to all three charts: 
Data sources: Global Carbon Budget (2025), Population based on various sources (2024) – with major processing by Our World in Data. 

2. Rewrite the header title and header subtitles to be insight-driven and relevant to the over-arching narrative.

3. For the Uneven Progress slope chart, add a footnote clarifying that the left axis shows each country's individual peak year, which varies by country.

For the final polish, open each chart in the Flourish editor and use the Assistant for changes that are easiest to spot once the chart is rendered with real data: overlapping annotation labels, tooltip formatting, spacing, legends or small layout adjustments.

You need to apply these edits chart by chart, depending on what each visualization needs.

When you’re happy with all three charts, open one of them in the Flourish editor and select Create a story. This takes you into the Story editor, where you can add each chart in sequence.

From there, choose how the story should work: keep the default format and add captions between charts, or switch to autoplay mode for a more presentation-style experience. Read more about stories in our help doc.