Skip to main content
Loading...
Skip to article
  • Qualtrics Platform
    Qualtrics Platform
  • Customer Journey Optimizer
    Customer Journey Optimizer
  • XM Discover
    XM Discover
  • Qualtrics Social Connect
    Qualtrics Social Connect

Using and Editing Your Brand Tracker Data Source


Was this helpful?


This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.

The feedback you submit here is used only to help improve this page.

That’s great! Thank you for your feedback!

Thank you for your feedback!


About Using and Editing Your Brand Tracker Data Source

BX programs collect data on your brand in addition to competitive brands and the larger market, which causes the dataset to be more complex than standard projects. BX programs use a stacked dataset (Brand Tracker Data Source, or BTDS) to identify insights in your data more easily.

The brand tracker data source in the projects list

Understanding the BX Data Source

Attention: The Brand Tracker Data Source is created as an Imported Data Project in the BX Program.

The BTDS varies from what you’d see in a standard dataset. In a standard dataset, each respondent has one row that contains all of the answers to their responses, with each brand’s metric as its own column. These datasets tend to be very wide with hundreds of columns.

Example: In this flat dataset, there is one row of data for Sarah Smith and one row of data for Barnaby Jones. There are separate columns for the metrics for each brand.
An example of a standard dataset, with one row for each respondent

In the BTDS, Brand becomes a first-class column in the dataset and every respondent has a row for each brand. The brand’s row contains all of the data for that single brand. These datasets have more rows than the standard dataset, but they have much less columns, which make them easier to read.

Example: In this stacked dataset, there are 5 rows of data for Sarah Smith, even though they represent only 1 response. Each row contains Sarah’s data for each brand measured.
An example of a stacked dataset, with three brand rows for each respondent
Qtip: Since stacked datasets have different rows for each brand, the Recorded Responses count in the BTDS Data & Analysis tab will be higher than the total number of unique respondents. To find the total number of unique respondents, create a filter for Singular = 1.

RESPONSEID

The ResponseID field helps us identify which rows belong to the same respondent. This value is from the original survey submission and is repeated for every row belonging to that respondent.

Highlighted the response ID column

SINGULAR

The count of Recorded Responses in the Data & Analysis tab will give us the total count of all rows, which will be higher than the total number of unique respondents. To determine the unique respondents, we can filter for the field Singular.

Highlighted the singular data column

  • When Singular = 1, we are shown unique respondent rows with no brand information. There is one unique respondent row per response.
    Qtip: Creating a filter for Singular = 1 will show the number of individual respondents.
  • When Singular = 0, the row contains brand data. There are multiple rows of brand data for each respondent.
Attention: You should thoughtfully consider whether to filter Singular = 1 or Singular = 0 depending on your data goals in dashboards. In most cases, you want to filter for Singular = 0 to exclude the extra non-brand rows from your base size and prevent data inaccuracy. Alternatively, if you need to show true counts of respondent-level data, you would filter for Singular = 1.

UNIQKEY

The __uniqKey__ field combines the ResponseID and brand name, shown as BrandName:ResponseID. For the unique respondent row with no brand information, the same row where Singular = 1, the __uniqKey__ will be :ResponseID. This is useful for narrowing down the exact response this data is from, as well as the brand it specifically gives feedback on.

highlighted the uniqkey column

BRAND

The Brand field shows which brand the data in that column is referring to, which allows you to easily see and filter for brand data.

highlighted the brand column

ATTRIBUTE-LED MULTI-SELECT DATA

Attribute-led questions contain the brands as the answer choices. These are typically multi-select question types, and there are multiple columns with data corresponding to that question.

Example: A common example of these question types is the “Aided Awareness” question.
The aided awareness question, with a list of brands as the answer choices

three fields highlighted for aided awareness question

  1. Question: Determines if the brand was selected for that question. The data will be a 1 (“Yes”) or 0 (“No”).
  2. Question – Displayed: Determines if the brand was displayed for that question. The data will be 1 (displayed) or 0 (not displayed).
  3. Question – Selections: Contains all the brands that were selected for that question. There will be multiple values in this column, and the value list will be the same for all rows of that respondent.

LOOP AND MERGE QUESTION DATA

There are also multiple columns for questions that pipe in the brand name using loop & merge.

Example: An example of this question type is the “Future Intent” question.The future intent question, with the loop and merge piped text highlighted

The two future intent columns

  1. Question: Shows what answer choice was selected for the question.
  2. Question – Displayed: Determines if the brand was displayed for that question. The data will be 1 (displayed) or 0 (not displayed).

NON-STACKED DATA

Non-stacked data are not related to brand, such as standard questions (e.g. demographics) and non-stacked embedded data fields. These fields are repeated per brand, which keeps it the data available whether you’re looking at one specific brand or across all brands.

The age data column highlighted, as an example for demographic data

Qtip: Multi-select questions using non-brand data will contain the list of choices selected in the same column. No information is stored on whether or not the respondent saw the answer choices.

Generating the BTDS

If you create a BX program from scratch, the Brand Tracker Data Source is not generated automatically and must be generated before data collection. Any data collected prior to the BTDS being generated will not be stacked.

Qtip: We recommend generating the BTDS after your survey is fully programmed and tested. Generating it earlier could end up complicating the dataset with extra material and columns.
  1. Create a BX Program.
    BX program, with the brand tracker survey highlighted
  2. Navigate to your BX survey.
  3. Enter the survey flow.
    survey flow with embedded data element
  4. Find the __bx_schematized__. It should be the first embedded data field in the survey flow.
  5. Change the value to 1.
  6. Click Apply.
  7. Navigate back to the survey editor and click Publish to publish your survey.

Stacked Dataset Compatibility

Stacked datasets have limitations on which field types and question types work best for data processing. While creating your BX program, it is important to ensure the survey structure is compatible with the BTDS that will be generated.

The BTDS compatible with the following question types:

  • Multiple Choice (single and multi-select)
  • Text Entry, Matrix
  • Descriptive Text
  • Constant Sum
  • Slider
  • Rank Order
  • Meta Info
Attention: Using incompatible question types could cause your data to stack in unexpected ways, or to not stack at all.

Filtering and Exporting the BTDS

Filtering and exporting the BTDS works the same as filtering and exporting from the Data & Analysis tab. These operations can be helpful to better understand dashboard insights, narrow the dataset size, or view specific subsections of the stacked dataset.

Qtip: Filtering for Singular = 0 will refine the dataset to only include unique responses for the BX program. This response count should match the data in the program’s survey project.

Optimization

BX programs with large brand lists can create datasets that are larger than necessary. With brand optimization, you can limit your data to only the brands that are significant.

Example: Let’s say you are doing research on brands in both the United States and Canada. There are 10 brands of interest in the US while there’s only 4 brands of interest in Canada. The Canadian research team would use optimization to isolate their 4 brands and not have to work with unnecessary data.
  1. Navigate to your BX project’s Survey Flow.
  2. In the embedded data block, click Add a New Field.
    Embedded data block with add a new field option
  3. Name the field
    gbt__opt-v1__Brand.
  4. Click Options.
  5. For the variable created in step 3, select Multi-Value Text Set as the variable type.
    Field type drop down
  6. Click OK.
  7. After the question that defines the brand list, add a branch. In the example above, this branch would be after the question about countries.
  8. Click Add a condition.
    Add a condition button highlighted
  9. Create the condition that will define your brand list. For the example above, this would be “Canada”.
    Branch with a condition created, then an embedded data element with the optimizer field
  10. Add the variable that you created in step 3.
  11. Enter the brands that you would like to report on. For the example above, this would be the 4 brands that Canada is interested in.
  12. Repeat steps 7-9 for the other brand lists that you would like to optimize.
  13. Click Apply.
  14. Publish your survey.

With the optimization variable defined, the processed BTDS will remove rows that are not included in the list for that variable.

Troubleshooting with BTDS

Attention: If any changes are made to the survey after a response has been collected, the BTDS will need to be reprocessed in order for those changes to be applied to existing responses. To reprocess your BTDS, contact Qualtrics Support.

Common issues with the BTDS include:

  • The response count is different between the BTDS and the survey’s Data & Analysis.
  • The brand questions or fields are not stacking or stacking improperly after the BTDS is generated.

RESPONSE COUNT

To compare response counts between the survey dataset and the BTDS, filter the BTDS for Singular = 1. Compare this count to the total count in the survey dataset. If data is flowing correctly, these two numbers should match.

If these numbers do not match, there could be incompatible components in the survey. Review BX Programs Best Practices. If all components appear correct, please contact Qualtrics Support with a list of affected response IDs.

NOT STACKING OR STACKING IMPROPERLY

Qtip: For more information on stacked dataset programming, see BX Program Best Practices.
  • Check the Reusable Choice list and all brand-led questions. The brand names must match up exactly to the Reusable Choice list.
  • Ensure there are no substrings within the brand list (e.g., one brand “Qualtrics” and another brand “Qualtrics Employee Experience”).
  • Ensure the question text is an exact match for each question that is meant to stack together. Different piped text fields in each question (e.g., piped text for different logos per brand) will prevent questions from being stacked together, which is especially common in Matrix questions.
  • Check the Expert Review > Data Stacking results to see if there are any flagged issues.

FAQs