← All Kits · Power BI Kit

Choosing a Power BI Visual: Nineteen Options, Six Questions

Michael Nocito · Updated August 2026 · Every number on this page was worked before it was published

The visuals pane has more than thirty entries before you install anything, and a good report uses about six. The extra choice does not improve reports; it mostly produces gauges.

What you do: choose from the comparison the reader has to make, exactly as you would in any tool, then use the Power BI specific ones only where they earn their place.

The short version. Card, bar, line, table, and a slicer. Everything else needs a reason.

Six questions, six answers

That last row deserves more respect than it gets. Users export to Excel constantly, and a well-built matrix with sensible subtotals is frequently the thing they came for.

The Power BI specific ones worth knowing

Matrix. Rows, columns and values, with expandable hierarchies and subtotals you can switch off per level. Conditional formatting on a value column turns it into a heat table, which is often the densest honest way to show a lot of numbers.

Small multiples. Available on bar, column and line. One panel per category, same axes, and it replaces most of the reasons people reach for a dual axis chart. If you are about to build eight similar charts by hand, use this instead.

Decomposition tree. Genuinely useful for exploring why a number is what it is, one split at a time. It is an exploration tool rather than a reporting one: excellent in a review, poor in a page somebody scans daily.

Key influencers. Runs a model behind the scenes and reports which factors are associated with an outcome. Treat every output as a correlation, because that is what it is. See correlation vs causation before quoting it in a meeting.

The four to avoid

Maps deserve a sentence on their own. Use one when the geography is the finding, not when the data merely has a country column. A map that shows the population distribution of your customers is usually showing the population distribution of the country.

Custom visuals from AppSource

1. Prefer certified visuals. Certification means Microsoft has reviewed the code and the visual cannot call external services.
2. Check whether your tenant even allows uncertified visuals; many do not, and a report built on one will simply not render for the audience.
3. Ask who maintains it. An abandoned visual becomes a broken report at the next platform update.
4. Ask whether a core visual plus formatting gets you 90 percent of the way. It usually does.

Formatting that matters more than the visual choice

1. Sort deliberately. The default is often alphabetical, which is almost never the useful order.
2. Turn off the data-point limit surprises by aggregating in the model, not in the visual.
3. Set number formats in the measure, not per visual, so every appearance of a number matches.
4. Give every visual a title that states the finding, as in chart titles.
5. Use the same colour for the same thing on every page of the report.

How to apply this to your own work

1. Count the visual types on your busiest report. More than six is usually a page doing several jobs at once.
2. Replace any gauge with a KPI card and see whether anybody misses it.
3. Look for eight near-identical charts and replace them with small multiples.
4. Check every visual for a Top N filter you did not set.
5. Ask three users which visual they actually use. It is usually the table.

The one habit to keep

Write the sentence the visual should produce, then pick the visual that makes that sentence easiest to reach. The pane is a menu, not a plan.

How many different visual types are on the page you maintain most?