> ## Documentation Index
> Fetch the complete documentation index at: https://docs.make-pretty.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Charts and data

> Pretty builds real PowerPoint charts wired to your numbers, not pictures of charts.

Ask for a chart and you get a native PowerPoint chart object. Click it and the ribbon's **Chart
Design** and **Format** tabs light up. Right-click a classic chart and **Edit Data** opens the
numbers behind it. Nothing about it is an image.

When someone else picks the deck up, they can change a figure, add a series, or restyle the
chart without you, and without Pretty.

## What it can build

| Family         | Types                                                |
| -------------- | ---------------------------------------------------- |
| Column and bar | Clustered, stacked, 100% stacked, 3-D                |
| Line and area  | Plain, with markers, stacked, 100% stacked, smoothed |
| Circular       | Pie, doughnut, exploded, pie-of-pie, bar-of-pie      |
| Numeric axes   | Scatter (markers, lines, smoothed), bubble           |
| Sequential     | Waterfall, funnel                                    |
| Hierarchy      | Treemap, sunburst                                    |
| Statistical    | Histogram, Pareto, box and whisker                   |
| Other          | Radar, stock, combo charts, region map               |

Chart chrome is available too: axis formatting, data labels, legends, titles, per-series and
per-point styling, error bars and trendlines. Ask for it in words ("put data labels on the last
series only", "start the axis at 60, not zero") rather than describing PowerPoint menus.

<Note>
  A **region map** chart shades geography by value. If you want a map you can recolour district by
  district as editable shapes instead, that is the [Maps asset](/assets/maps), which is a
  different thing.
</Note>

## Handing over the data

<Tabs>
  <Tab title="In the prompt">
    Fine for anything you can type. Give the series names, the categories and the units.

    ```text theme={null}
    Column chart of revenue by region for FY24 and FY25, in EUR millions:

    EMEA      41.2   52.8
    Americas  63.7   68.1
    APAC      18.9   29.4

    FY24 in grey, FY25 in the accent colour, data labels on FY25 only.
    Title: "APAC is where the growth is."
    ```
  </Tab>

  <Tab title="From a file">
    Attach a `.csv` or `.xlsx` and name what to plot. Pretty reads the workbook, including
    which sheet and which columns.

    ```text theme={null}
    From the attached actuals.xlsx, sheet "Monthly": plot net revenue by
    month for 2025 as a line chart, with a second line for the same months
    in 2024. Ignore the rows below the "Total" line.
    ```

    Attach the file by dragging it onto the pane (it shows **Drop to reference in chat**) or
    with **Upload files**.
  </Tab>

  <Tab title="From a document or screenshot">
    Pretty can pull figures out of a PDF's text or read a table out of a screenshot.

    ```text theme={null}
    Table 3 on page 11 of the attached annual report has segment revenue for
    the last three years. Build a stacked column chart from it, one column
    per year, one segment per series.
    ```

    A scanned PDF with no text layer comes back empty, since there is no OCR. Screenshot the
    table instead and attach that.
  </Tab>
</Tabs>

<Warning>
  Be explicit about units and scale. "Revenue 41.2" could be thousands, millions or a percentage,
  and a chart that silently picks the wrong one is worse than no chart. State the unit and put it
  in the axis title.
</Warning>

## Prompts that produce good charts

Say what the chart has to prove, not just what to plot. The takeaway drives the chart type, the
sort order and what gets emphasized.

```text theme={null}
Waterfall showing how we got from 12.4m opening ARR to 15.1m closing ARR:
new business +4.2, expansion +1.6, downgrade -1.1, churn -2.0. Positive
steps in the accent colour, negatives in red, totals in dark grey.
```

```text theme={null}
Scatter of our top 30 accounts, contract value on the x axis and support
tickets per month on the y. Label only the five accounts above 50 tickets.
The point is that our biggest accounts are not our noisiest ones.
```

```text theme={null}
100% stacked bar of headcount by function for each of the last four
quarters, so the shift towards engineering is visible. Four bars, one per
quarter, engineering at the bottom of every stack.
```

## Restyling a chart you already have

Click the chart on the slide so it is selected, then ask. The selection badge in the composer
tells Pretty which chart you mean.

```text theme={null}
Recolour this chart to the master palette, drop the gridlines and the
chart border, move the legend to the bottom, and set the axis labels in
the deck's body font at 10pt.
```

```text theme={null}
Same chart, but sort the categories by FY25 descending, grey out
everything except the top three, and add a dashed trendline to the
second series.
```

Changing the numbers works the same way. Say what changed and Pretty updates the chart's data
rather than rebuilding it from scratch:

```text theme={null}
Q4 came in at 71.3, not 68.1. Update the Americas FY25 value and leave
everything else alone.
```

## If the chart is not right

* **Numbers look off.** Right-click a classic chart and choose **Edit Data** to see exactly what
  it plotted. That is the source of truth, not the drawn chart.
* **Too busy.** Ask for fewer series, a sorted order, or labels on one series only. "Remove
  everything that is not carrying the argument" is a valid instruction.