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

Key Drivers Widget


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 Key Drivers Widgets

The key drivers widget allows you to see the correlation between one outcome metric and one or more potential drivers.

Key Drivers Widget

Types of Dashboards

This widget can be used in a few different types of dashboard. This includes:

Field Type Compatibility

Only Number Sets, Numeric Values, and individual items from field groups (CX) and categories (EX) are compatible with the key drivers widget. For more information on field types and widget compatibility, check out our Field Type & Widget Compatibility Tables.

Widget Customization

Outcome Metric

The outcome metric is a measure of progress that is influenced by key drivers. For example, a company might be concerned about their clients’ overall satisfaction with a particular product or service. The outcome metric in this case would be an overall satisfaction score.

Outcome Metric option in left editing pane

Potential Drivers

Potential drivers are performance-based metrics that influence the outcome metric. For example, if a company’s outcome metric is overall satisfaction for a product or service, potential drivers might include quality, value, or usefulness.

Potential Drivers section in left editing pane

Specifying Field Bounds of Drivers

You can specify field bounds for your drivers when your widget metric is set to either average or top / bottom box. Depending on the selected metric, the setup is different.

metric field in editing pane

If you would like to specify the upper and lower absolute limits of a field value (so the widget knows how to make its calculations), you can specify the field bounds for any potential driver that you add to your widget if the metric is average. You will click on the name of the potential driver, check the box for Specify Field Bounds, and then set your field minimum and field maximum for the selected driver. The purpose is to allow the bounds to be greater in both directions than what is observed in your response data.

Example: For example, you may have a multiple choice question where respondents can select choices 1-10, but only choices 2-7 have been selected thus far. If you’d still like to make your calculation based on the highest possible choice, you can do so by specifying your minimum and maximum field bounds as 0 and 10, respectively.

Specify Field Bounds within Potential Driver menu

If you would like to specify the upper and lower absolute limits of a field value and the metric is top / bottom box, you can do so by specifying the box range. You will click on the name of the potential driver and then move the sliders for the Box Range to set your field minimum and field maximum for the selected driver.For Top Box Bottom Box, adjust the sliders

Qtip: The values come from your selected field’s recode values.

Performance Axis

The performance axis refers to the x-axis of the key drivers widget.

Label and Threshold Marker in Performance Axis options

Changing the Threshold Type will adjust the vertical line along the x-axis:

  • Static: Determine where the vertical threshold line will lie on the x-axis. Moving the Threshold Marker allows you to decide the point at which a score changes from performing well to performing poorly.
  • Dynamic: The threshold line will automatically be set to the median values of the drivers being pulled into the widget.
    Qtip: Customer satisfaction data is often driven by the data collected, not hardcoded standards. In cases where there are no industry standards, this option can be favorable.

This axis can be renamed by typing your desired name in the Label text box.

Importance Axis

The importance axis refers to the y-axis of the key drivers widget.

Label and threshold itself in Importance Axis

Changing the Threshold Type will adjust the horizontal line along the y-axis:

  • Static: Determine where the vertical threshold line will lie on the y-axis. Moving the Threshold Marker (not pictured) allows you to decide the point at which a score changes from performing well to performing poorly.
  • Dynamic: The threshold line will automatically be set to the median values of the drivers being pulled into the widget.
    Qtip: Customer satisfaction data is often driven by the data collected, not hardcoded standards. In cases where there are no industry standards, this option can be favorable.

This axis can be renamed by typing your desired name in the Label text box.

Display options

Select from the different Display Options to further customize the widget.

Select Show X Axis to display the Performance percentages along the bottom of the widget.

Show X Axis checkbox in Display Options section

Select Show Y Axis to display the importance values along the left side of the widget.

Show Y Axis checkbox in Display Options section

Select Show Labels to show the labels next to the drivers within the widget.

Show Labels checkbox in Display Options section

Select Show Legend to display the legend in the widget. Click on the color swatch to change the color of the driver circles for each quadrant. You can also select the default text and type in your own legend values.

Show Legend options in Display options menu.

Select Scale Range Automatically to adjust minimum and maximum axis values automatically. This does not adjust your threshold markers. Rather, it serves to “zoom in” or “zoom out” to give you the best possible view of your key drivers.
Scale Range Automatically checkbox within Display Options section

Select Scale Data Points by Sample Size to adjust the size of each driver circle relative to the other driver circles’ sample sizes. The larger the circle, the larger the sample size.

Scale Data Points by Sample Size checked in the Display Options

Show number of responses for each data value in tooltip ensures that when someone hovers over a data point, a tooltip will show them the performance, importance, and sample size for that data point.

Qtip: This setting can only show the total responses for up to 10 data values at once. After 10 data values, the tooltip will only show the number of responses for the value you are hovering over.

Hovering over dot on the key drivers widget, tooltip shows the numbers mentioned

Interpretation

The Y-axis, also called the importance axis, is a value between 0 and 1 that represents how strongly a given driver is correlated with the outcome metric. It is calculated by taking the absolute value of Pearson’s r, such that:

Importance = | r |

As the importance value gets closer to 1, the relationship between the driver and outcome is understood to be stronger.

The X-axis, also called the performance axis, is a normalized scale. This means the value ranges from 0% to 100%. This axis is normalized, and depending on whether you selected average or top / bottom box for your metric, is either dependent on the average score or the top / bottom boxes of scores. Normalizing makes it possible to compare potential drivers with different scales. The percentage for average is calculated by taking the value for the outcome metric’s potential driver and dividing it by the maximum possible value of the potential driver.

Example: Let’s say you ask respondents to answer a question on a scale from one to five. If the highest score a participant gives for this question is a four, then four will be used as the denominator when calculating the percentage on the performance axis.

The percentage for top / bottom box is calculated by taking the value for the outcome metric’s potential driver and dividing it by 100.

The key drivers widget is divided into four quadrants:

  • Important and highly rated: These values fall in the top right quadrant and indicate drivers that play a large role in determining the outcome measure.  These drivers also have higher scores. For example, “Service Satisfaction” drives overall satisfaction in such a way that higher perceived safety is related to higher overall satisfaction scores.  In this case, respondents have also indicated that this company is performing well with regards to “Service Satisfaction.”
    Important but Highly rated in top-right quadrant of graph
  • Important but poorly rated: These values fall in the top left quadrant and indicate drivers that play a large role in determining the outcome measure. However, these drivers have lower scores. For example, “Product Satisfaction” plays a big role in determining overall satisfaction in such a way that poor products are related to lower overall satisfaction scores. In this case, respondents indicated that this company is not doing well in regards to controlling “Product Satisfaction.” This is an area of improvement for this company.
    Important but Poorly Rated in top-left quadrant of graph
  • Not important and poorly rated: These values fall in the bottom left quadrant and indicate drivers that are not important in determining the outcome measure. These drivers also have low scores. For example, “Frequency of Contacting Support” doesn’t drive overall satisfaction scores and respondents also indicated this company was not having to contact support often. However, this company might not need to improve on this driver because it isn’t affecting their overall satisfaction.
    Not Important and Poorly Rated in bottom-left quadrant of graph
  • Not important but highly rated: These values fall in the bottom right quadrant and indicate drivers that are not important in determining the outcome measure. These drivers also have a high score. For example, “Resolution Time” does not drive satisfaction, but it was given high scores by respondents. While one might argue that having a good resolution time for escalated issues is always a good thing, it does not influence this company’s overall customer satisfaction.Not Important but Highly Rated in bottom-right quadrant of graph

Analysis Method

The key drivers widget has two different analysis methods that can be applied to it. This affects how the values are displayed in the four quadrants and the importance score. You can change the calculation used in your widget at any time – just keep in mind that the importance score and the values in the four quadrants may change.

Qtip: Pearson Correlation is more performant than Relative Importance for large datasets. When the total dashboard response count across all data sources exceeds 5 million, Relative Importance analyses may time out or fail.

selecting the analysis method in a key driver widget

  • Pearson correlation: This calculation measures the linear correlation between the drivers and the outcome metric.
    Qtip: For all key driver widgets created before February 18, 2022, the default and only calculation was Pearson Correlation. All widgets created before this date have been left as Pearson, but you can now switch the calculation if desired. Alternatively, creating a new key drivers widget will enable access to the newest version of this functionality.
    Qtip: Measure groups can only be used as an outcome metric when Pearson correlation is selected.
  • Relative importance analysis: This is the default calculation method. Relative Importance analysis runs models with every combination of the independent variables to determine the independent impact on the r-squared (variance explained), because all the drivers are likely to be highly correlated. Please note that when you first set up your widget with Relative Importance analysis, it may take some time before it loads and displays data. To learn more about this method of analysis, also called Johnson’s Relative Weights, see our Stats iQ page.
    Attention: Relative Importance analysis is not currently available in EX dashboards.
    Qtip: Relative Importance analysis does not currently support using measure groups as the outcome metric. Additionally, you cannot use org hierarchy dashboard filters with Relative Importance analysis.
    Attention: Relative Importance analysis requires a complete response count greater than the number of drivers. A complete response is one that contains data for every single driver. Relative Importance analysis supports up to a maximum of 50 drivers at a time.

Deciding Between Calculations

Pearson Correlation does not account for multicollinearity, which is when the independent variables (in this case, drivers) being compared to the dependent variable (an outcome metric) are already highly correlated themselves. Relative Importance, in contrast to Pearson, does consider multicollinearity. That’s why it is the industry standard for survey data.

Example: Let’s say you’re trying to use whether a customer’s heard of a given brand and their preference for a given brand to predict their purchase intent. Are customers who haven’t heard of your brand very likely to indicate it as their top preference? It’s unlikely; naturally, the less aware of a brand a customer is, the less likely they are to show a preference for that brand. This is the problem multicollinearity presents. Because of this, it is important to tease out those relationships between drivers in order to determine which will have the greatest impact on the outcome metric.
Attention: Relative importance analysis is only available in CX Dashboards.

Additional Options

Click on the three horizontal dots in the top-right corner of your widget to view additional options.

the three dot menu in the top right corner of a widget

  • Export: Export your widget as a JPG, PDF, or XLSX.
  • View data: View a table containing data on all the key drivers within your widget.Key drivers data points table

FAQs