When to use a bar chart

When a bar chart is the right answer

When your data has discrete categories — labels, regions, products, periods — and one number per category. When you want the reader to compare those numbers easily. When the number of categories is small enough that they all fit on one chart (roughly 3 to 30). When the encoding "longer means more, shorter means less" matches your story.

When something else is the right answer

Use a line chart for trends with many time points. Use a histogram for distribution of one continuous variable. Use a scatter plot for the relationship between two variables. Use a heatmap when you have two categorical dimensions and one value. Use a sankey or treemap for hierarchical part-of-whole. Use a table when readers will compare specific numbers more than the overall pattern.

Borderline cases

Time-series with under 20 points: bars work, but a line works too. Pick whichever your audience expects. Many categories with skewed values: bars work, but consider a log scale or splitting into "top 10 + Other". Two values per category over time: bars work as a grouped chart, but a slope chart can be more elegant for two periods.

Genres bar charts handle especially well

Rankings (top-N anything). Survey results (counts of answers). Budget breakdowns (spend by category). Performance reviews (scores by competency). Quarterly metrics (one number per quarter). Election results (votes by party). All of these have the same shape: discrete categories, one number each, where comparison is the point.