Stepthrough BI Start free

What is aggregation in BI and Tableau?

Quick answer

Aggregation is combining many rows into a single number for each group, using a function such as SUM, AVG, MEDIAN, MIN, MAX, COUNT or COUNT DISTINCT. "Total revenue by region" is revenue aggregated with SUM, grouped by region.

Last updated

Updated · By Robert Breen

Why it matters

Charts show groups, not rows, so every measure on a chart is aggregated. The function you choose quietly decides the answer. SUM, AVG and MEDIAN of the same column can tell three different stories.

Knowing which one to use, and saying so in the chart title, is one of the quickest ways to earn trust in your numbers.

A worked example

Five deals closed this month: $2,000, $2,500, $3,000, $3,500 and $40,000. SUM is $51,000. AVG is $10,200. MEDIAN is $3,000. A sales manager asking "what does a typical deal look like?" wants the median; the average is pulled up by the one big deal.

A second trap is COUNT vs COUNT DISTINCT. If the data has one row per order line and 300 orders have 780 lines, COUNT(Order ID) says 780 orders and COUNTD(Order ID) says 300. Only the second is right.

Often confused with

Aggregation vs filtering

A filter removes rows before anything is calculated. Aggregation combines the rows that remain.

Aggregation vs a Make aggregator

In Make, an aggregator bundles many items into one for the next step of a workflow. Same idea, different job: it moves data rather than summarizing it for a chart.

Aggregation vs a pivot table

A pivot table is a spreadsheet tool for doing aggregation. BI tools do the same thing every time you put a measure on a chart.

Tips

  • Use median for "typical" values with outliers (deal size, time to close).
  • Use COUNTD on IDs whenever rows are not one per thing.
  • Label it: "Median deal size", not "Deal size".

Learn it in a free lesson

Related in the AI glossary

Frequently asked questions

What does ATTR mean in Tableau?
ATTR returns the value when every row in the group has the same value, and an asterisk (*) when they differ. It is common in tooltips.
Should I use average or median?
Median when a few large values could skew the result, such as salaries, deal sizes or response times. Average when values are fairly even, or when you need it to add up to the total.
Why does my total look doubled?
Usually a join that repeats rows, or counting order lines as orders. Check with COUNTD on the ID.

All business intelligence terms, A to Z · Free business intelligence lessons