You create a metric definition in Tableau Pulse by connecting to a published data source on Tableau Cloud. This data source should have time series data that is complete, well-structured and clearly labelled. Follow the guidelines in this article to structure your data source. For information about creating metric definitions based on a data source, see Create Metrics with Tableau Pulse.

General data source requirements

Make sure that the data source you’re working with meets the following criteria.

• It’s a single published data source. You can't connect to a data source that is embedded in a workbook, and you can't connect to multiple data sources or use data blending, unless you combine the data before publishing the data source. The data source can be an extract or a live connection, and it can use a virtual connection or connect directly to the data.
• You have the Connect, View and Create Metric Definitions permission capabilities for the data source.
• The data source contains:
  • A measure to be aggregated as a sum, average, median, maximum or minimum or a dimension to be aggregated as a count or count (distinct).
  • A time dimension for the metric's time series. Tableau Pulse monitors data over time, so single point-in-time values won’t produce a valid metric. The granularities supported for the time series are day, week, month, quarter and year. Data that requires a lower level of granularity (hour or minute) isn’t a good fit for Tableau Pulse.
  • At least one dimension that can be used to filter the data and insights.

Tableau Pulse displays the dimension names and values that you add as filters on metrics and in insights text. These names and values in your data source should be easy for others to understand. To create the best experience for viewers, follow these recommendations so that the names and values fit with the conversational language of Tableau Pulse.

• Use title-style capitalisation rather than all caps.
• Use spaces rather than hyphens or underscores to separate words.
• Don’t use special characters.
• Don’t use Boolean fields that appear as true or false.
• Don’t use fields that consist of IDs or other indecipherable strings.

Custom calendars, goals and thresholds require additional data to be added to your data source. For information about setting up your data source for those features, see Requirements for custom calendars and Requirements for data source goals and thresholds.

Requirements for custom calendars

Set a custom calendar in Tableau Pulse if your organisation doesn’t use a standard Gregorian calendar. For example, you can define a calendar that aligns with the National Retail Federation 4-5-4 calendar(Link opens in a new window).

A custom calendar must be defined in your data source, and all of the calendar fields must belong to the same table. The custom calendar doesn’t need to belong to the same table as your metric data. If the custom calendar is in a separate table, create a relationship between that table and the metric data table.

Don't add row-level security or data source filters that remove dates from the calendar table. If you add a filter to the transactions table in your data source, make sure that the filter scope is set to only that table and not that table and related tables. For information about setting the scope for data source filters, see Create a data source filter.

Each row of the custom calendar represents a date and contains these fields.

• Reference date
• The standard Gregorian date that you're mapping to the custom calendar
• Data type: date
• Example: 01/01/2025
• Day of year
• Data type: number (continuous dimension)
• Example: 333
• Week of year
• Data type: number (continuous dimension)
• Example: 48
• Month of year
• Data type: number (continuous dimension)
• Example: 11
• Month name
• Data type: string
• Example: December
• Quarter
• Data type: number (continuous dimension)
• Example: 4
• Year
• Data type: number (continuous dimension)
• Example: 2024

The format for dates in Tableau varies based on the locale that a user has set, and some fields aren’t used in all locales. For example, locales set to Japanese or Chinese show the ‘month of year’ instead of the ‘month name’.

After you have added the custom calendar to your data source, you configure the calendar as part of the metric definition. For more information, see Set a custom calendar.

Custom calendar validation

Make sure that the custom calendar is complete, consistent and that it doesn’t skip any dates or numbers. Follow these guidelines to ensure that your custom calendar is valid.

• All of the fields are present for each reference date.
• The fields are of the correct data type.
• The first entry in the table is the first reference date.
• The reference dates are in chronological order.
• The reference date includes every day of the year, even if there’s no business data for that day.
• The day of year, week of year, month of year and quarter of year begin with 1 and increase by 1 with no skipped numbers.
• There are 7 days in a week. If a year ends with a partial week, the next year must begin with a partial week such that the 2 weeks add up to 7 days. For example, if week 53 of a year has only 2 days, then week 1 of the next year must contain 5 days to form a 7-day week, and week 2 would contain days 6–12.
• The month name is the same for each month number across all years in the calendar.
• There are 4 quarters in a year.
• Each quarter starts and ends with a complete month. Months can’t be split across quarters.
• The data contains the 2 previous calendar years. Metrics with a comparison to the prior year use those dates to determine the correct comparison.
• The data is complete through to the end of a custom year. If future data goes into a new custom year, include the complete set of dates for that year, but don’t include more than 4 years beyond the end of the current custom year.

Tableau Pulse validates the fields in the custom calendar. If there’s a problem with your custom calendar, you’ll see an error that helps you troubleshoot the calendar.

Example custom calendar

To understand how to structure your custom calendar, download an example custom calendar. The table below shows ten rows from the example custom calendar.

Requirements for data source goals and thresholds

There are two ways to set goals and thresholds: manually or as part of the data source. A manual goal or threshold is simply a static number entered for an individual metric, so there are no data source requirements. A data source goal or threshold is set at the definition level. These goals or thresholds use a field from your data source to show a dynamic value on the metrics based on that definition. For more information, see Configure data source goals and thresholds and Set Manual Goals and Thresholds with Tableau Pulse.

If you want to include a data source goal or threshold as part of a metric definition, there are additional data source considerations.

• Include the field that specifies the goal or threshold values as part of the data source that the definition connects to. Metric definitions connect to only a single data source, so you can’t connect to a separate data source for this data.
• If the goal or threshold data is in a separate table from the core definition data, create a relationship between the two tables. If you join the tables instead of relating them, the data source might end up with duplicate goal or threshold values. For information about relating data, see Use Relationships for Multi-table Data Analysis.
• If you want a finer-grained goal or threshold to be shown when users apply dimensional filters on metrics, include that level of granularity in your data. Otherwise, Tableau Pulse displays the total goal or threshold value for a dimension, even if the data is filtered to a subset of values.

These two tables show example data sources with goals of different granularity. If there isn’t goal or threshold data for each row, for example in the case where there is only a monthly goal, include that value only once. Don’t duplicate the value in each row.