If you want to see how well a template teaches Copilot, ask it for a slide with data. Titles and bullets tend to come out acceptably on a decent template. Charts and tables are where generated decks most visibly drift: default colors, cluttered axes, a legend nobody needs, a title that names the chart instead of saying what it shows.
That’s not because Copilot can’t make a chart. It’s because charts are new objects, created from scratch, and they only look like your brand if the template has told Copilot what your brand’s charts look like. This post covers how.
Why data slides drift
Three things determine how a generated chart or table looks:
- The theme — chart colors come from the theme’s accent colors, in order. If your brand colors are painted onto existing shapes but aren’t in the theme, generated charts use defaults.
- Styles — chart, table and shape styles defined in the template. Microsoft recommends applying consistent styles for “color schemes for charts and data visuals,” icons, shapes and image placeholders.
- Examples — sample slides showing how your organization presents data. Microsoft’s guidance asks for layouts for “highlighted statistics and key takeaways” and examples of charts, tables, processes, timelines, diagrams and infographics.
Most templates handle the first partially, the second rarely, and the third almost never.
Step 1: Get chart colors into the theme
The theme has six accent colors, and charts use them in sequence: the first data series takes Accent 1, the second Accent 2, and so on. So the order of your accents is a data-visualization decision, not just a branding one.
A few principles:
- Put the color you want for “the main series” in Accent 1. In many charts, one series is the story and the others are context.
- Make adjacent accents distinguishable. Two similar blues in Accents 1 and 2 make two-series charts hard to read.
- Consider a neutral accent. A gray in the palette lets examples demonstrate “highlight one, mute the rest” — one of the most effective patterns in business charts.
- Check contrast. Every accent needs to read against your background colors, and text on colored bars needs to read against the bars.
Step 2: Define the styles
Build one chart of each type your organization commonly uses — column, bar, line, pie or donut if you must, and any others you rely on — and format each the way your brand presents data:
- gridlines light or removed;
- axis labels in the body font, at a readable size;
- data labels where they help, instead of forcing readers to trace to an axis;
- legends only when there is more than one series and labels can’t do the job;
- no 3D effects, shadows or gradients unless your brand genuinely uses them.
Do the same for tables: header row treatment, row banding or dividers, alignment of numbers (right-aligned, consistent decimals), font sizes. Tables in generated decks often inherit PowerPoint’s default table style, which is rarely anyone’s brand.
These formatted charts and tables then become the basis of your example slides.
Step 3: Build data example slides
This is the step that teaches behavior, not just appearance. Examples should cover the data patterns your people actually present:
Chart with a takeaway title. The most important single habit in business charts: the slide title states what the data shows (“Northern region drove most of this year’s growth”), not what the chart is (“Revenue by region”). Showing this consistently in examples is one of the most effective lessons you can teach.
Highlighted statistic. One big number with a short explanation. Microsoft calls these out explicitly. They are among the most common and most effective data slides, and generated decks often bury the number in a bullet instead.
Comparison. Two or three values side by side, with the difference made obvious.
Trend. A line or column chart over time, with the relevant change emphasized.
Table. A clean, readable table at the density your organization actually uses — including a heavier example if reference tables are common.
Process and timeline. Not strictly charts, but data-like structures that generated decks frequently get wrong.
For each, use illustrative, clearly generic numbers — round figures that obviously aren’t real. Generative models can reproduce example content closely, and a realistic-looking statistic in an example slide can surface as a false fact in someone’s deck. Label example charts in a way that couldn’t be mistaken for real data.
Step 4: Put the rules in writing
Examples show; rules tell. Since September 2026, brand managers can publish Brand Skills — written rules in Brand Kit — and write note instructions in template speaker notes. Data rules that tend to be worth writing down:
- “Chart titles state the finding in a full sentence.”
- “Include a source line under every chart and statistic.”
- “Highlight the most important series in the primary brand color; show others in gray.”
- “Use a highlighted-statistic slide when one number is the main point.”
- “Never invent or estimate figures not present in the source material.”
That last rule is worth special attention. Test it specifically: give Copilot source material with gaps and check that it doesn’t fill them with plausible numbers.
Step 5: Test with real data
Build a set of test prompts from realistic source material — a spreadsheet summary, a quarterly report, a messy set of figures in an email — and check:
- Appearance: brand colors, fonts, styles; no defaults.
- Choice: the right kind of chart for the data.
- Emphasis: the important number or series is the most visible.
- Titles: findings, not labels.
- Accuracy: every figure matches the source. No invented numbers.
Accuracy is the one no template can guarantee. Microsoft’s guidance is that Copilot output should be reviewed by a person, and data slides deserve the closest review of all.
Key takeaways
- Generated charts and tables are new objects; they only look like your brand if the theme, styles and examples teach them.
- Order theme accents deliberately — charts use them in sequence.
- Define chart and table styles once, and build examples on them.
- Teach takeaway titles, highlighted statistics and clean tables through examples, using obviously illustrative numbers.
- Reinforce with written data rules, and check every figure in review.
Want your data slides to look like yours? Have your template scored.**
Related reading
- Theme colors vs. hardcoded colors: the setting that decides your output
- Designing example slides a model can learn from
- Light, medium, heavy: designing for content density
Sources
- Microsoft Support, Keep your presentation on-brand with Copilot (July 2026)
- Microsoft Support, Manage template settings in your Brand Kit for Copilot in PowerPoint (August 2026)
- Microsoft 365 Message Center, MC1473166 — Brand Skill support (16 Sep 2026)