Impact allows an organization to easily examine the relationship between engagement driver scores and other key metrics at a segment level. Amongst other issues, this can help them understand the return of investment in engagement within an organization. Impact metrics work across all Peakon question sets, whether it is engagement, diversity, and inclusion, health and wellbeing, or COVID-19.
This article will cover:
Introduction to Impact
After importing a metric into Peakon, Impact metrics will display a scatter plot with a trend line between these values. An impact score will be provided, displaying how a business metric KPI will change as the associated driver score changes, as well as an explanatory power value explaining the strength of this relationship.
It is always recommended to lead analysis with a research question and to think of an expected outcome first. From this, you can consider how the outcome of the analysis meets or contradicts your initial ideas and drives the narrative of your investigation.
Examples of research questions:
- Is there a link between increasing employee wellbeing and reducing average absence days per employee?
- Do teams with higher workload scores also perform higher on productivity metrics?
- Do locations with higher engagement scores also have lower first aid incidents?
- Are more engaged teams providing better customer service?
- Are departments with lower growth scores more likely to have higher employee turnover?
Impact Metrics can be used for any metric, but these are some general use cases:
|Attrition Rate %
|Project Completion %
|Sales Growth %
|Customer Retention %
|Other Driver Scores
|Customer Satisfaction %
|Sales Achievement %
|Mystery Shop Scores
How it works
The statistical technique used in this tool is linear regression. It is a method for estimating the relationships among predictors and outcomes by fitting the best linear (straight line) relationship. The regression will output the following measures:
The impact value
The impact value is the regression coefficient that tells you the direction and effect between your predictor and outcome variable. Specifically, the impact value will show the expected change in your metric with a one-point change in your Peakon driver score.
What is a good impact value?
It will largely depend on the type of metric you are regressing your Peakon driver score against. For example, when examining the relationship between attrition and engagement, it would be expected that the impact value is a negative score due to more engaged employees being less likely to leave their roles.
The explanatory power
The explanatory power, also known as the R-squared value in statistics, is used to evaluate how well the regression model fits the real-world data you have imported. While the impact value explains the relationship between a predictor and outcome variable, the explanatory power explains to what extent the variance of one variable explains the variance of the second variable.
What is a good explanatory power?
According to academic research, an explanatory power value of 12% or below indicates low, between 13% to 25% values indicate medium, 26% or above, and above values indicate high model fit. A good rule of thumb is that the higher the explanatory power, the better the model explains changes in your outcome variable. These boundaries may seem low, but when you consider the multifaceted nature of many metrics - employee engagement is often just one factor out of many contributing to a metrics success.
How to improve result validity
To improve the validity of the findings it is recommended to replicate the analysis often. Finding the same trends and effects over different time periods is an indication that the findings are robust and not coincidental.
- Provided data needs to be granular i.e., come from a higher number (e.g. greater than 30) of smaller units.
- All the units/segments need to be independent i.e., contain data from different employees, no overlapping segments.
Administrators or leaders that have the Administer Impact metrics permission enabled can add a metric that is any measure commonly used in their company, and it can be a percentage, percentage of increase/decrease, or a positive or negative raw value.
The metric needs to be for a group in your company that corresponds to a segment in Peakon. The feature may be used for all Peakon segments with the exception of the Manager segment.
How to upload the metric
- Go to Administration > Impact
- Click on Add metric
- Give the metric a name
- Select the parent attribute of the segments you'd like to upload
- Click on Next step
- Download the template with the pre-formatted fields
- Fill in the fields as required and import the file into Peakon
This will import the values to the correct segments as specified in the spreadsheet. Once uploaded, close the sidebar and the uploaded data will be visible in a table. Impact metrics are now ready to view in the Analysis section.
How to edit a metric
- Go to Administration > Impact
- Choose an existing metric
- Click on Edit metric
It is possible to edit the name, description, driver, minimum/maximum values, the relationship, the activity status, add new data points and delete the data set entirely.
Impact metrics are available in the Analysis area, and they are available to anyone with the Impact permission enabled under the Access Statistics section within an Access Control group.
The information included in the Impact graph is contextual and will therefore show the relationship based on the context view that the user is in. The context is visible in the top left corner, within the context switcher box.
On the Impact page, the view can be customized by using the available filters. For example, the below graph shows the relationship between the engagement score and attrition rate %. Hover over a data point to visualize the segment.
It's also possible to export your Impact analysis to Microsoft Excel.
The next graph shows the relationship between the health and wellbeing score and attrition.
The graph will always return to the default settings upon leaving the page.