#Tableau Prep26 discussing

Hi, @Mathias Berg

I’m not sure if this is what you were looking for, but please take a look. 

It seems that before long, Tableau will allow you to link data sources together. 

You can prepare separate data sources for each object and then combine them. 

By splitting up the data processing, you may be able to shorten refresh times. 

Since this is a long‑awaited feature for many users, it’s exciting to see it coming. 

<tableau help> 

https://help.tableau.com/current/pro/desktop/en-us/datasource_extend_compose.htm

Hi @Viviane Dornelas

 I wasn't able to open up the Tableau Prep file you contributed, and I was going to submit a packaged Tableau Prep File in return but found a number of issues in the data that need to be addressed before doing so.   

I'd recommend the use of Tableau Prep in advance of Desktop based on what you're trying to do, and because of the need to update the data regularly and because of the below. 

(1) Will the 'Report (the one with Publication Dates)', not the weekly report data, also be regularly updated/added to the original? If yes, then that too should be a wildcard union (in addition to the weekly report data, also being a wildcard union). 

(2) In reviewing the Output File attached to your post, there are a number of duplicates.  For example:

The differences being Recipient Partner (there are two) and Publish Date there are two additional for each timestamp.  

In addition, it appears that the weekly data being a union is actually adding in prior values, so each weekly report is actually reflecting just changes in the data, 

and likely you only need to use the most recent weekly report data (maximum report date) joined to the publication report data - if that makes sense

? 

For example, parsing down to just Seabord recipient and just unique document ID CDE-EU2-DO5002-000021 there are 36 rows (if adding back in the other recipient, would be 72 rows of data):

 As we can see in the most recent data for June/2026 are highlighted at bottom, and we can see in the previous data it is almost the same data preceding June's data, meaning that changes occurring during each ''weekly report" within the totality of the report.  I also see revision letters of A or B which may also help with fine-tuning the duplicates down to just 3 rows per each recipient per date type. 

 Additionally, in the 'Publish Date Report' the separate report that contains the Publication date, there can be be multiple dates AND times, which is also contributing to the issue of duplication.  Is there a business rule for keeping either the minimum or maximum publication date/time per each recipient?  That would help bring down some of the duplication I'm seeing.  For example Document ID Number 14:

There are four different timestamps - which one to use?  You'll know better than us as to what you need to use or not use out of the duplicate rows we're seeing.   

Let us know what you think about each of these concerns? 

This behavior is a known limitation with Tableau Prep + Tableau Cloud when using “Append to Table” with certain connections like SFMC.

Root Cause

• Append / Incremental refresh requires a stable, owned published data source with valid run-as identity
• With SFMC connections: 
  • The connection does not fully persist user context/credentials for append operations
  • Tableau Cloud fails to resolve the run-as user, resulting in: 
    • “No Tableau Server user found”
    • Blank datasource reference
• “Create Table” works because it creates a new datasource each run, avoiding this dependency

Why this happens

• Append requires: 
  • Existing published datasource with consistent owner + embedded credentials
  • Proper connection mapping during flow run
• SFMC connector + embedded credentials often does not support append/incremental reliably in Prep flows

Correct and practical approaches

Option 1 (Recommended):

• Avoid Append
• Use Create Table (full refresh) instead

Option 2:

• Publish the target datasource separately
• Ensure: 
  • Credentials are embedded and valid
  • The datasource owner is explicit and unchanged
• Then reference it in Prep (still may fail due to connector limitation)

Option 3:

• Use an intermediate storage: 
  • Write to Snowflake / DB / Hyper
  • Then handle append from a supported source

Key Conclusion

• This is not a permission issue
• It is a connector + Tableau Cloud limitation with Append/Incremental on SFMC
• Full refresh (Create Table) is the only consistently reliable approach

Bottom line:

Append and incremental refresh with SFMC in Tableau Prep Cloud are

@Matthew Laforest

came across your post trying to discover other ways to increase load time between Salesforce and Tableau Prep. Similar to what Diego shared, the best way I've found to increase load times from Salesforce in Prep is to use the Custom SQL option and query only the fields you need, and filter down in that step. It does use SOQL rather than SQL so there are some limitations there. However, when connecting directly to the Salesforce table in Prep, even if you filter down the input step, it still queries every single row and column despite the filter. Going the Custom SQL route actually does narrow the query down so I've seen much faster load times going that route. 

If you've found any other solutions I would love to hear them if you're willing to share! My super large queries are still taking way too long for my liking so I'm trying to find anything else I can implement for faster loading times.

First, can you confirm that TabPy is actually running and reachable from the Mac? For example, if TabPy is running on the default port (9004), try accessing the endpoint from a browser or using a command such as:

curl http://localhost:9004/status

If that does not return a response, the issue is likely with TabPy itself rather than Tableau Prep.

A few other areas worth checking:

• Verify that Tableau Prep and TabPy are both running natively on the same architecture (Intel vs Apple Silicon), especially if you're using Python environments created on a different platform.
• Confirm that macOS Firewall or security settings are not blocking local connections.
• Check whether Tableau Prep is configured to connect to localhost, 127.0.0.1, or a hostname that resolves correctly on your Mac.
• Review the Tableau Prep logs and TabPy logs for connection-related errors. They often provide more detail than the UI.
• Make sure the TabPy version is supported by the version of Tableau Prep you are using.

@Jose Baltazar

Do you mind to try with Tableau Prep Builder 2026.1.0: 

It seems Tableau Cloud Prep has been updated to 2026.2.0 (This Tableau Prep Builder version has not been released):

Thus, if you can't open the flow, mostly this can be related to the update. 

If you are not able to edit the flow (browser of 2026.1.0 release) then I recommend you to create a support case: 

If this post resolves the question, would you be so kind to "Accept this Answer"?. This will help other users find the same answer/resolution and help community keep track of answered questions. Thank you. 

Regards, 

Diego Martinez 

Tableau Visionary and Tableau Ambassador 

Hi @Shannon S. We read our entire Salesforce task table into Tableau Prep or Desktop frequently. Maybe you have more tasks. By default, Salesforce only keeps 12 months of tasks in the database even though you see more on-screen. You have to get Salesforce support to extend that period, we went to 24 months which slows everything down but is more comprehensive. 

In Tableau Prep you can remove fields and add filters at the connect step to save memory.

In Tableau Desktop you can filter using Custom SQL/SOQL.

https://developer.salesforce.com/docs/atlas.en-us.soql_sosl.meta/soql_sosl/sforce_api_calls_soql_select.htm

So you can add a WHERE clause to filter

SELECT Id, Name FROM Zip_Assignments__c WHERE Name='77630'