FAQ: People Analytics

What is Storyteller?: Storyteller is an automated analytical engine that:
Searches millions of combinations of data.
Makes connections between the combinations of data.
Surfaces the most significant results in the form of stories.
What is augmented analytics?: Augmented analytics is an approach that uses statistical techniques and advanced analytics to automate findings in the data that aren’t easily attainable by doing manual analytical work. Storyteller brings the principles of augmented analytics into practice.
What is a story?: A story is a narrative summary of a finding in the data that answers a business question. Example: What are key turnover trends? Each story describes an instance of significant under or overperformance in a metric. To enable better decision making, a story also provides the associated trend and a list of the most important drivers contributing to the metric.
How are stories generated?: Storyteller:
Groups the input data into different views by aggregating metrics for each dimension and different combinations of dimensions.
Filters out any views that have no data or are statistically insignificant.
Makes connections between the views to create a network that defines the relationships between views. This network of relationships provides context when evaluating 1 view compared to another. Example: The view of
Location = Chicago
has these relationships:
The parent of
Location = Chicago; Cost Center = Sales
.
The child of the whole organization.
The sibling of
Location = New York
.
A proxy to
Cost Center = Sales
because the entire Sales department is in Chicago and 90% of the workforce in Chicago are salespeople.
Analyzes the views in the network of relationships to determine their significance.
Returns the most significant views in the form of stories.
How does Storyteller determine the most relevant information?: Storyteller analyzes and evaluates the statistical significance of each view of the data. It clusters views together based on several factors, including:
Commonality of the underlying data.
Hierarchical relationship.
Similarity of problem.
It then ranks each view based on the greatest impact to the related business question and displays the most significant views as stories.
What kinds of analysis do the stories use?: Each story is the result of either:
Trend analysis: Storyteller compares the current month metric performance in a dimension or combination of dimensions to the historical performance.
Gap analysis: Storyteller compares the current month metric performance for a dimension or combination of dimensions to the internal peer group average. Example: If the dimension is supervisory organization, performance is compared to Company or Level 1.

Depending on the story, the population might be:
All active workers.
All terminated workers.
All active and terminated workers.
A subset of those populations filtered by Level 1 in either the Primary Hierarchy or Secondary Hierarchy.
How can we compare metric performance across dimensions in stories?: In order to normalize the metric performance of any organization or region and make them comparable, Storyteller weights performance by the underlying employee population. This helps reduce the outsize impact of large percentage changes in a small population.
To do this normalization, Storyteller:
Calculates the average of the metric for the combination of dimensions in the story.
Calculates the average of the metric for the relevant population. Example: Company or subset containing peers.
Calculates the difference or delta between these 2 averages.
Weighs the delta by multiplying it by the size of the population of the combination of dimensions.

Storyteller tells you how much the metric for that combination of dimensions needs to change to be consistent with the metric for the total population.
Why do some top drivers contribute more than 100%?: When a top driver contributes more than 100% to the negative performance of a metric in a story, other factors might have an opposite effect that contributed positively to the performance.
Example: If a story displays a red negative number, the top drivers for that story also display as red negative numbers. Workday ranks the drivers by their contribution to the underperformance of the metric. On the
Drivers
tab of the
View More
dialog, a corresponding green bar displays if there’s a red bar that’s over 100%.
How do the stories change as time progresses?: All stories are recalculated each time Storyteller runs. After each snapshot update, there's a new set of stories displayed on the dashboard based on the current data. In all these cases, what drives the display of stories is a statistical ranking of the over or underperformance detected. In some cases, changes to the configuration or the underlying algorithms might also affect what stories are displayed.
It's possible that a story in a given dimension, such as Org, Mgt Level, or Region, appear again after the monthly data refresh. The repetition implies that the performance gap or trend identified in the previous month remains statistically significant and ranks among the top stories for the business question in the current month. Users can still view performance numbers that include the most recent monthly update.
How does the scoring method for stories work?: The scoring algorithm takes these factors into account:
Statistical significance of each story. Is the detected change of the trend or deviation from population average important enough?
Value of a story. What is the calculated impact on the business question?
Relationship to the other stories. Is this story a driver of another story, the effect of another story, or part of other similar stories?
Why do some stories have only 2 top drivers and others 10?: Workday curates the list of dimensions for analysis depending on the business question to ensure that Storyteller provides the most relevant analytical slices. Within those dimensions, Storyteller displays the top 10 drivers that significantly contribute to the metric performance in a story.
How does cardinality impact People Analytics?: Cardinality defines the size of the population in the dimension, directly impacting what stories are surfaced as well as story quality and actionability. Smaller populations don’t provide interesting and meaningful insights, which can result in unnecessary iterations of the mapping process. A high cardinality can result in long computation time of the stories.
How can I view more information about a particular business question or metric?: You can click the information icon next to each business question or metric to view more information, such as:
Why a business question is important.
The metric that drives the business question.
The calculation behind a metric.
Why do some metric trends display in a neutral color?: The expected result for select metrics can vary across organizations, as some metrics don't follow a universally accepted trend.

Example: To establish parity in gender equality, an increase in female representation might be the expected direction for organizations that have a low percentage of females, while a decrease might be the focus for organizations where female representation is already at 70%.

To address differences in metric trends across organizations, we set a target range for these metrics:
Average Span of Control for values between 5-15.
Female Representation for values between 40-60%.

We display movement within these ranges in a neutral color. Movement towards these ranges displays in green, while movement away from these ranges displays in red.

All changes for the metric Average Target Compensation display in a neutral color, as individual cases can determine whether an increase or decrease is expected.

Concept: Storyteller Engine