What do I do if I think my data in my Tableau visualization is wrong?

What do I do if I think my data in my Tableau visualization is wrong?

If you see inconsistencies in your data while viewing your Tableau workbook, please consider going through Alight Analytics’ QA process before sending a ticket to the Data Solutions team.

Step 1: ISOLATE THE PROBLEM

First and foremost, figure out which visualization within your dashboard looks off, and make a copy of the worksheet that contains that visualization. Your troubleshooting should occur within the new tab. Don’t mess with the work you’ve already done to try to find an issue.

Step 2: CHECK YOUR EXTRACTS

Once you have navigated to the worksheet, you’ll need to check your extracts. This is because we want to make sure we are working with the most up-to-date data. In the top left corner of your worksheet, check and see if the data source you are using is extracted. The image should look like this:

If you are not using an extract, please proceed to Step 3.

If you are using an extract, first check the extract filter by clicking on the data source tab and clicking on the "edit" button in the top right. If that does not work, navigate to the worksheet with the error, right click on the data source and uncheck “Use Extract”.

If unchecking did not fix your issue, please move forward to Step 3.

Step 3: CHECK YOUR FILTERS

If you have any filters in your visualization, try removing them. If you absolutely need filters to identify where the problem is, show the data in tabular form and show the filter as a dimension instead.

Check to see if your filter is applied to multiple or all visualizations in your dashboard. Consider removing the universal filter first and then isolate the problem visualization as outlined in Step 1.

Try editing your filter. New values may have been added to the data set that were not available when the filter was created.

    

Your filter could be from a secondary data source or a filter may be applied at the connection level (there are cases where this is actually needed but it is somewhat rare).

Step 4: CHECK YOUR BLENDS

The issue may be within your blends. To check, first, build a table from each data source and include all blended dimensions in each table. Once complete, compare your 2 new tables with the data from your original visualization.

Did you get the same numbers?

If so, your data inconsistency is probably within the data itself. Check the other options outlined in this tutorial like filters or extracts next.

If you did not get the same numbers as your original visualization shows, your issue is most likely with the blend. Remember: Blends work like a left join. If there are any values in the secondary data source that don’t exist in the primary data source, those rows will be lost in the blend.

Also make sure that you are blending on similar fields.

Pro tip for blending Date vs. Date & Time with potentially unavailable time series data: If you are blending Google Analytics data (secondary) with Media data (primary) and media doesn’t run on some days, those days of GA data will also be lost in the blend. This can be rectified by creating a new date field that’s more aggregated (to the week, month, etc.) and blending at that level.

Step 5: IS IT TABLEAU?

Check out this link on Tableau’s order of operations and see if this may be impacting the data returned.

Step 6: IS THE PROBLEM WITHIN YOUR ONEVIEW/MEDIA VIEW?

This should be the absolutely last step.

Create a table in a new worksheet with the impacted data source(s) within your OneView/Media View. For example, if your AdWords conversions aren’t right in OneView/Media View, connect to the AdWords data set that’s being used for the OneView/Media View and check the conversions total there.

If the individual view matches what you are seeing in the OneView/Media View, your issue is likely with the data.

If it does not match your OneView/Media View, there is likely an issue with your OneView/Media View logic and you will need to open a ticket with Data Solutions to investigate further.

Was this article helpful?
0 out of 0 found this helpful
Have more questions? Submit a request

Comments

0 comments

Please sign in to leave a comment.