# Minimum granularity for data source design

*Tableau Consultant Study Guide · Plan and Prepare Data Connections · Last updated November 3, 2026 · https://www.certifyforce.com/study-guide/tableau-consultant/plan-and-prepare-data-connections/minimum-granularity-for-data-source-design*

Find the coarsest grain that still answers every requirement, and know what pre-aggregation destroys: finer date levels, averages over the original rows, distinct counts, medians, security attributes and incremental refresh.

- Certification: Salesforce Tableau Consultant
- Exam section: Plan and Prepare Data Connections
- Study guide: https://www.certifyforce.com/study-guide/tableau-consultant

**Official exam objective:**

> Specify the requirements for minimum level of granularity.

## What the exam expects

- Derive the minimum grain of a data source from a set of requirements, keeping every dimension that can't be derived from another.
- Predict which questions a summarized table or aggregated extract can no longer answer.
- Keep the attributes row-level security filters on when the data is summarized.
- Deduplicate a snapshot table to one row per entity in Tableau Prep.
- Keep a small extract refreshing incrementally by summarizing upstream instead of in the extract.

Key terms:
- **Grain:** what one row represents, such as one order line or one store per day.
- **Minimum grain:** the coarsest grain from which every requirement can still be aggregated.
- **Non-additive aggregate:** an aggregation whose results can't be aggregated again along a dimension, such as COUNTD or MEDIAN.
- **Roll-up:** storing dates at a coarser level, such as Month, in an aggregated extract.
- **Entitlement:** a combination of attributes the data can be filtered on for security.

## Key facts

| Fact | Detail | Source |
|---|---|---|
| Recommended data shape | As granular as possible rather than aggregated | [Data tips](https://help.tableau.com/current/pro/desktop/en-us/data_tips.htm) |
| Non-additive aggregates | COUNTD, MEDIAN, RAWSQLAGG: calculated individually, not re-aggregated | [Troubleshoot Data Blending](https://help.tableau.com/current/pro/desktop/en-us/multipleconnections_troubleshooting.htm) |
| Aggregate data for visible dimensions | Measures use their default aggregation; dates can roll up to Year, Month and so on | [Extract Your Data](https://help.tableau.com/current/pro/desktop/en-us/extracting_data.htm) |
| Hidden fields | Excluded from the extract | [Extract Your Data](https://help.tableau.com/current/pro/desktop/en-us/extracting_data.htm) |
| Incremental refresh | Unavailable with Aggregation; requires All Rows | [Refresh Extracts](https://help.tableau.com/current/pro/desktop/en-us/extracting_refresh.htm) |
| Dynamic user filter | Data must include the security information | [RLS options](https://help.tableau.com/current/server/en-us/rls_options_overview.htm) |
| Prep rank default order | Descending; add ASC to change | [Prep calculations](https://help.tableau.com/current/prep/en-us/prep_calculations.htm) |
| ROW_NUMBER | No skipped values; not in the visual calculation editor | [Prep calculations](https://help.tableau.com/current/prep/en-us/prep_calculations.htm) |

## Check yourself

- When can a ROW_NUMBER deduplication return a different surviving row on each flow run?
- Why does AVG over a summarized table differ from the average of the original events?
- Which RANK variant gives tied rows the rank of the last instance of the value?

## Sources

- [Data Aggregation in Tableau](https://help.tableau.com/current/pro/desktop/en-us/calculations_aggregation.htm)
- [Extract Your Data](https://help.tableau.com/current/pro/desktop/en-us/extracting_data.htm)
- [Tips for Working with Your Data](https://help.tableau.com/current/pro/desktop/en-us/data_tips.htm)
- [Troubleshoot Data Blending](https://help.tableau.com/current/pro/desktop/en-us/multipleconnections_troubleshooting.htm)
- [Refresh Extracts](https://help.tableau.com/current/pro/desktop/en-us/extracting_refresh.htm)
- [Connect to a Custom SQL Query](https://help.tableau.com/current/pro/desktop/en-us/customsql.htm)
- [Create Level of Detail, Rank, and Tile Calculations](https://help.tableau.com/current/prep/en-us/prep_calculations.htm)
- [Aggregate, Join, or Union Data](https://help.tableau.com/current/prep/en-us/prep_combine.htm)
- [Overview of Row-Level Security Options in Tableau](https://help.tableau.com/current/server/en-us/rls_options_overview.htm)
- [RLS Best Practices for Data Sources and Workbooks](https://help.tableau.com/current/server/en-us/rls_bestpractices.htm)
- [How Analysis Works for Data Sources that Use Relationships](https://help.tableau.com/current/pro/desktop/en-us/datasource_multitable_analysis_overview.htm)

## About the full unit

The "Core concepts", "Exam traps" and "Worked scenario" sections of this unit, its 7 linked practice questions, and unlimited timed mock exams are included with the Tableau Consultant pack (USD $19): https://www.certifyforce.com/practice-exams