Table Insights highlight potentially noteworthy or unusual areas in your company by leveraging descriptive statistics to add information and visual cues to Storyboard tables.
Jump to the video  One AI Table Insights
Some scenarios where Table Insights apply are as follows:
 Do you have a high amount of attrition in a certain area of the company?
 Are certain teams promoting a disproportionately small (or large) percentage of people?
 Does a gender imbalance exist anywhere in the company?
Descriptive statistics are useful in answering the questions above because both the rate and population size to which that rate applies matter. A team of 4 people where 2 of those people terminate employment is a different scenario than 500 people leaving from a team of 1000. The methodology leveraged by Table Insights factors in both rate and population size and applies greater “weighting” to larger populations.
The visual cues applied by Table Insights include sorting, highlighting, and filtering; all of which are configurable in both view mode and design mode in Storyboards. The result is a table where key information can be easily identified.
What Information Do Table Insights Provide?
Table Insights enrich tables meeting certain criteria in Storyboards by adding the following:
 The overall average rate for the entire population contained in the table.
 A Pvalue column added to the table indicating magnitude of difference between each group and the overall population.
 The ability to sort the table by the pvalue, in order from the lowest value (greatest magnitude of difference from the overall population) on top to the highest value.
 The capability to filter the table to only display potentially noteworthy values.
 The ability to apply highlighting to rows where the difference between the group rate and the overall rate exceeds the configured threshold, in either a positive or negative direction.
 Configuration of the Alpha threshold, meaning cutoff below which a pvalue is considered noteworthy.
 Configuration of Tailedness, which is the direction in which the group rate differs from the overall average.
 An information icon in the tile header that when selected describes how Table Insights work.
 An explanation of the pvalue for an individual group that is provided if a pvalue is selected.
How Do Table Insights Work?
At the present time, Table Insights are designed specifically to work on tables containing a proportional rate and a population size broken out by one or more dimensions. The statistical method leveraged by Table Insights is called a “one proportion ztest”. This test was chosen because it works well for proportional rates. For a technical explanation of the one proportion ztest, this is a good article to check out: https://www.statology.org/oneproportionztest/.
Table Insights highlight areas in your company where proportional rates are unusual. This is achieved by comparing a rate and population size for each group in a dimensional breakout (the sample) to the overall rate and assigning a probability value, or pvalue, to each group. There’s a lot packed into that short description so let’s break it down:
Proportional rate
A proportional rate metric is one in which a subset of a population is divided by that entire population.
Overall rate
The overall rate is the average rate for the entire population.
Population size
The population size is the number of individuals in the group or overall population and also the denominator of the rate metric.
Pvalue
The pvalue measures the probability that the differences between each group and the entire population is likely due to chance. Pvalues are expressed as numbers between 0 and 1, where values close to 0 indicate a difference worth noting and values close to 1 suggest no noteworthy difference. In simple terms, pvalues help us decide whether the patterns we observe in our data are meaningful.
Please note that because as population size increases pvalues decrease, large companies will see more groups highlighted as noteworthy. It may be necessary to adjust the Alpha in some situations to compensate. See the "Table Insights Configuration" section below to see how to perform this.
In summary, pvalues help us make informed decisions by quantifying the likelihood of differences between groups being due to chance. Please note that Table Insights is not providing statements of statistical significance. It is rather a tool that leverages descriptive statistics to highlight groups that might be unusual or noteworthy.
Table Setup and Data Requirements for Table Insights
Table Insights currently apply to proportional rates broken out by one or more dimensions. By “proportional rate” we mean a subset of a population divided by that entire population. More specific requirements are as follows:
 The visualization type must be a list (table).
 The table must contain a proportional rate metric. The metric must be formatted as a percentage. Examples:
 Termination Rate
 Promotion Rate
 Headcount  % Female (EOP)
 The table must contain the population metric that’s the denominator of the rate metric. Examples:
 Average Headcount
 Headcount (EOP)
 At least one dimension (to make your groups) must be included.
 The table cannot contain any pivoted dimensions. This includes time dimensions.
 Rates of less than 0% or greater than 100% cannot be present for any included group.
Here is an example of an Explore query for Table Insights:
If you build your list in accordance with the instructions above, pin it to a Storyboard, and have the appropriate permissions, a lightbulb icon will appear in the upper right corner of the Storyboard tile. From this icon you can run Table Insights.
Permissioning Table Insights
Any user, regardless of their permissions, can view Table Insights configured to run automatically by a user possessing the permission to perform this configuration. All users are also able to disable Table Insights in Storyboard view mode using the Turn off Table Insights option in the tile menu and reenable it using the Run Table Insights option in the same menu.
The ability to run or configure new Table Insights, either on demand or in Storyboard design mode, requires that the user is assigned to an Application Access Role with the CanEnableOneAITableInsights permission checked.
Using Table Insights
Running Table Insights on Demand
Provided a properly structured table and the correct user permissions, Table insights can easily project results onto the tile in real time. The options to run or turn off Table Insights can be found by clicking the lightbulb icon in the upper right corner of the tile.
If the light bulb icon is not present on the tile, either you don’t have the application access necessary to run Table Insights or the data contained in the table does not meet the requirements.
Running Table Insights Automatically on Storyboard Load
Storyboard designers can enable Table Insights on tables as the default behavior. This means that whenever a particular Storyboard is opened the table will automatically includeTable Insights. The option to enable this feature is available in the Tile Settings on the Discover tab.
Table Insights Configuration
Table Insights can be configured in either view mode or design mode in Storyboards. In view mode, the configuration settings can be found by selecting the gear icon from the subheader of a table where Table Insights has been enabled.
In design mode, the settings are located in the Tile Settings on the Discover tab after enabling Table Insights.
The only differences between the settings in view mode and design mode are:
 In view mode settings are only applied for the existing session. In design mode they can be saved so they are automatically applied each time the Storyboard is loaded.
 Highlighting colors can be modified in design mode but cannot be modified in view mode.
While the settings are described on the configuration screen, a deeper understanding of Alpha and Tailedness is useful in fully understanding Table Insights.
Alpha
The alpha is the cutoff below which a pvalue is considered noteworthy vs. due to chance or coincidence. The default value for the alpha is 0.05. This means that there is a less than 5% chance that the difference between this group's rate and the overall rate is due to chance.
Table Insights will produce a pvalue for each observed rate based on how different the observed rate is from the overall rate; the larger the difference, the smaller the pvalue. If the pvalue is less than alpha, we conclude that the observed rate is noteworthy. In statistics, Alpha is typically set at .05 or .01 by convention. The smaller the alpha value, the more confident we are that the observed rate is truly different from the overall rate, but smaller alpha values may result in failing to recognize truly noteworthy differences.
Tailedness
Tailedness is the direction in which the group rate differs from the overall rate.
 Selecting “Less Than” or “Greater Than” considers differences in a specific direction. This is also known as a “one tailed test”.
 Selecting “Either Direction” considers differences in both directions ("greater than" and "less than") and is used when you want to detect any noteworthy change, above or below the overall mean. This is also known as a “two tailed test”.
Please note that switching from “Either Direction” to “Greater Than” or “Less Than” will result in the pvalue numbers being halved.
Video  One AI Table Insights
Watch this video to learn about One AI Table Insights.
Please note that despite utilizing Pvalues, table insights does not provide statements of statistical significance.
You will learn:

What types of questions table insights can help answer.
 How table insights work behind the scenes.
 How to build the list queries that table insights can be leveraged with in Storyboards
 How to leverage the custom configuration options to create the clearest visual and data story with table insights.
Run time: < 10 minutes
Comments
0 comments
Please sign in to leave a comment.