Most Data Warehouse destinations has these options:
​

For file storage Destinations this means you will get one file per day, for BigQuery this means that each day is it's own partition. We will always only send parts that have been updated.

Use cases: When you want to export a lot of data. When you only want to reload the data for the specific day that has changed to decrease load costs in your pipeline, and maybe you have Data Sources that changes for specific dates far back in time like CRM Data Sources. When you or a tool you use prefer to store or load data as daily files. When you believe that your BigQuery table being partitioned per day will be more performant.

Limitation: The date range of the Data Share can include a maximum of 5 years of historical data. Data is not sent fully aggregated, see this article.

For file storage Destinations this means you will get one file per month, for BigQuery this means you will get the table partitioned per month. This will also mean that the entire month is sent whenever any data changes in that month.

Use cases: When you want to export a lot of data. When you prefer working with monthly files and that becomes a good aggregation level. When you don't want too many files or when you want to export data farther back than 5 years. When you believe that your BigQuery table being partitioned per month will be more performant.

Limitation: If one day in a month changes the entire month has to be re-exported. Data is not sent fully aggregated, see this article.

The date range of the Data Share can include a maximum of 20 years of historical data.

You will get one file will all the data, the data will also be fully aggregated. This option supports way less data amounts (max 1M rows) than the previous options.

Use cases: When you want to export metadata like campaign details to store separately or be used as lookup values. When you want to export a smaller aggregated report as a file or table. When you only want to export the last x days or last x weeks of data.

Limitation: You can export a maximum of 1 million rows in one Data Share.

When you have data sources without values for the built-in "Date" dimension in Funnel this is the only way to export that data directly. Note that usually you have non-daily data because you want to do a lookup in Funnel and then export it in a normal (daily, monthly or single) Data Share and then this is not the correct option.

Data not stored by day can for example be from Google Sheets imports when you select this option:

Use cases: When you want to be able to export your non-daily data directly.

Limitation: You can only export data that does not have values for Funnel "Date", you can not include "normal" data that has values for "Date" in these Data Shares.