Blending is such a powerful feature, especially with the new features of Tableau 8, but we've run into hard stops several times deep into a project and faced either a compromised-value outcome, or a painful data foundation change so blending is avoided. It would be extremely useful and open up another realm of intuitive flexibility if the following capabilities could be available on secondary data connections:
- Calculated Sets: when creating a calculated set (such as Top 10 Stores according to a measure) it would be very helpful to be able to choose a field which is from a secondary data connection, or is a calculated field which references fields from multiple data connections. Today, sets from secondary connections are grayed out, and calculated fields blending multiple connections don't appear in the list of fields available.
- Calculated Fields: currently any calculated fields which reference a secondary datasource must be aggregated; there are some cases where a secondary dataset is mandatory even though the data may share the same key structure as the primary, and we need to be able to do math at the detail level (sans aggregation in the calculated field) with fields from both data connections within the same calculated field.
- Ranking: similar to above, today only primary data connection fields are available to set ranking on a measure. Many times, it would cut through a lot of workarounds if it was possible to rank using secondary data connection fields, or calculated fields which cross multiple data connections.
- Sorting: while it's relatively easy to create a blended view, one simple frustration of blending is that sorting on a field from a secondary data source (particularly measures) is a one-time sort only. This isn't obvious until your data gets updated, and you notice your view is now in some wacky sort order. Typically you could set a sort on a dimension field through the Sort popup which contains a list of fields to choose from, but since only primary data connection fields appear, there's no clean way to automate this. A workaround we've used is to place a Discrete version of the secondary data connection measure in front of all the other dimension fields, and then hide the field. This would allow a 'live' Ascending sort, but further finagling is needed if you want a Descending sort. It would be so helpful to simply choose a secondary data connection field from the Sort dropdown in the popup.
- Mapping using Latitude/Longitude (generated): when blending data, it would be helpful if you could create a map which uses Lat/Long (generated) from primary datasource and/or secondary datasource. Today, generated Lat/Long fields gray out in the secondary data connection.
- Polygon/ordered line mapping: it would enable great flexibility for blending data and analyzing complex subjects if it were possible to draw polygons or lines using fields in a secondary datasource. Frequently we have two equally important data connections which both contain data which should be mapped as independent polygons (superimposed through dual axis), and we must simply choose one today. Today, the roadblock is that we can't drop a field from a secondary data connection into the Order shelf when using a Polygon or Line type map.
Last but not least... how about the ability to establish blending Relationships between secondary data connections?