Peakon’s Attrition Prediction identifies those segments in your organisation that have the highest risk of employees leaving. 

Attrition risk is shown at a segment level and displayed as either 'Minimal', 'Low', 'Reduced', 'Elevated', 'High', or 'Severe'. This allows you to take the appropriate steps to improve the segment scores and reduce the attrition risk.

Attrition Prediction uses trends observed in employee tenure and separation data.

This article covers the following sections:

  1. Factors that affect attrition risk in organisations
  2. How is attrition risk calculated?
  3. Viewing the attrition risk on segments
  4. Interpreting the scores
  5. Addressing attrition risk
  6. Improving the prediction accuracy
  7. Enabling attrition prediction

1. Factors that affect attrition risk in organisations

Employee attrition is a natural part of any organisation, however, our data shows some interesting findings in the engagement trends of those employees who voluntarily separate from organisations. 

An employee’s decision to leave is influenced by push and pull factors

When deciding to leave, employees weigh-up push factors against pull factors. Push factors are those which push an employee out of the organisation. These factors can be controlled by improving your organisation’s driver scores, in particular, those that are marked as priorities as these tend to have a higher impact on employee engagement. 

Pull factors, are those which an organisation has less influence over. For instance, job opportunities outside of your organisation. However, these factors can be actively mitigated in some cases. 

Employee engagement is highest during the first 24 months of tenure

On average, employee engagement declines by as much as one point during the first two years of employment, regardless of whether the employee has any intention of leaving. This is expected and the decline is likely to level-out after two years of tenure. 

It’s natural that employees are most engaged when they first start a new job. Ideally, employee engagement wouldn’t decline at all but the fact that it does should not be alarming. Employees who have many years of tenure in a company would also have gone through this process.

Employees are most likely to leave before reaching 24 months of tenure

Employees are at their highest risk of leaving an organisation around 3-12 months into their tenure. This is also when the biggest drop in engagement occurs - after 3 months of tenure. 

This presents a paradox as on the surface you will have an employee with higher engagement, relative to other tenure segments, but who is also more likely to leave your organisation. It’s therefore crucial to factor length of tenure into the attrition risk calculation as relying solely on engagement is not an accurate indication of attrition risk.

Employees who don’t participate in surveys are more likely to leave

Employees that have become disinterested in answering the survey have also become disinterested in trying to improve their workplace and therefore are at a higher risk of leaving your organisation. Improving the participation rate mechanically will not lower the attrition risk unless feedback is also actioned. If employee feedback is left unactioned it can actually lead to an elevated desire to leave. 

Therefore, it’s crucial to mobilise an active feedback loop between employees and your organisation. Adopting this method results in active survey participation being a strong predictor of lowered attrition risk.

Employees will tell you how loyal they are

Peakon’s standard ‘loyalty question’ (‘If you were offered the same job at another organisation, how likely is it you would stay at [company name]?’) is a strong predictor of intention to leave the organisation. Therefore, Peakon highly recommends enabling this question on surveys to increase the accuracy of Attrition Risk predictions. 

2. How is attrition risk calculated?

Peakon’s unique Attrition Risk model incorporates the following factors into the risk calculation shown on segments:

1. Each employee’s: 

  • Length of tenure 
  • Responses to the engagement question. If the engagement loyalty question (‘If you were offered the same job at another organisation, how likely is it you would stay at company x’) is in active use, this is also used. The loyalty question is a strong predictor of attrition so we recommend turning it on for higher accuracy.
  • Participation in surveys.

2. The model then utilizes engagement trends based on thousands of observation points from those employees who have been marked as “Resigned” using Peakon’s Separation date and Separation reason attributes.

3. The model will then compare the aggregated scoring pattern of this set of anonymized employees against the scoring pattern of those employees in your organisation.

4. Peakon then calculates the probability of each employee leaving your organisation within six month’s time to establish a company benchmark for attrition risk.

5. This score for each segment of employees is aggregated and the final attrition risk is displayed on the segment as either 'Minimal', 'Low', 'Reduced', 'Elevated', 'High' or 'Severe' compared to your company average.

3. Viewing the attrition risk on segments

The attrition risk score can be viewed on all sub-segments throughout the dashboard except for the Separation date and Separation reason attribute segments

The score will be either 'Minimal', 'Low', 'Reduced', 'Elevated', 'High' or 'Severe'. Hovering over the score on the segment will give a breakdown of the risk for a particular segment.

It’s also possible to view the attrition score from within the heatmap. Click on the icon at the top right of your screen to change the score type and select “Attrition risk”. The heat map will then display the risk for each segment.

The heat map will give display the following:

  • The attrition risk for each segment
  • The percentage of employees in each phase of the Employee Experience Cycle - this is based on data from the Tenure attribute (if in use) 
  • How many employees have resigned from each segment (this data is obtained from the Separation date and Separation reason attribute where "Resigned" has been selected, (if in use). This number will remain visible in line with the "Time for former employees" aggregation setting.
  • The aggregated participation rate
  • The latest round's participation rate
  • The engagement score
  • The loyalty question score (if the question is enabled)

4. Interpreting the scores

Peakon will give an attrition risk of either 'Minimal', 'Low', 'Reduced', 'Elevated', or 'High'. This reflects the attrition risk in a segment over the next six months, compared to the average risk in the rest of the organisation. The risk categories are calculated compared to the company’s overall attrition risk. This means that the categorization will take into account the attrition risk distribution in the company.

  • Minimal - Segments with Minimal attrition risk have an aggregated attrition risk in the bottom 10% compared to the rest of the company.
  • Low - Segments with Low attrition risk have an aggregated attrition risk in the bottom 25% compared to the rest of the company
  • Reduced - Segments with Reduced attrition risk have an aggregated attrition risk in the bottom 50% compared to the rest of the company
  • Elevated - Segments with Elevated attrition risk have an aggregated attrition risk in the top 50% compared to the rest of the company
  • High - Segments with High attrition risk have an aggregated attrition risk in the top 25% compared to the rest of the company
  • Severe - Segments with Severe attrition risk have an aggregated attrition risk in the top 10% compared to the rest of the company


Taking an example to illustrate how the model works. Imagine a small sales team of four -  three junior employees and one manager. This is a relatively new sales team with ¾ of the team having less than two years tenure at Kinetar Sales. 

The manager has three years tenure in the company and always completes the weekly Peakon survey. Two of the members joined 12 months ago. Their engagement, although high, has started to decline and one of them has stopped taking the time to complete the survey. The newest member of the team joined last month and has completed all the surveys.

On the surface, this team has a high engagement compared to the rest of the company. However, when you take into account this team’s length of tenure, and the fact that one team member has stopped participating in the weekly survey altogether, there is an elevated attrition risk compared to the company average. 

The manager has not been encouraged to share their dashboard with the team so they are not actively involved in the feedback loop. The Autonomy and Environment drivers have been a priority for this manager for some time now. The comments suggest the team find it difficult to concentrate at work due to lack of meeting rooms, meaning they are having to take meetings at their desk in a crowded, noisy office. They are frustrated that the company has not allowed them to work from home to alleviate the problem while a bigger office is being sourced. These are push factors, which Peakon helps you identify. 

One of the team members was contacted by a recruiter on Linkedin last week regarding an open position at a competitor. The job would mean better pay, less commuting time, plus they have a generous remote work policy - pull factors. This week, whilst contemplating whether he should apply for the new job, this particular employee decides to answer the weekly survey because the first survey question in the email caught his eye: 

“If you were offered the same job at another organisation, how likely is it you would stay at Kinetar Sales?”

Based on the low score given to this question, the attrition risk on the manager’s dashboard is updated in real-time from elevated to high, compared to the company average.

5. Addressing attrition risk 

The most effective way to address attrition risk is to improve the drivers of engagement. 

  1. Review the drivers that have been identified as priorities on your dashboard. 
  2. Click into each of them and review their associated comments. These will help you uncover the underlying issues. 
  3. Click on the improve tab within the dashboard for each of these drivers. 
  4. Review the suggested actions and recommended resources. These will give you ideas on how to improve your team’s feeling toward this driver. 
  5. Add some short and medium-term actions that directly address some of the concerns raised in the comments.
  6. Involve your team by sharing your dashboard and going through the results together and the actions you’ve created. This shows the team that their feedback is being acted upon and in turn drives participation and helps improve engagement.
  7. Review your Employee Experience Cycle reports. Pay particular attention to segments with less than 24 months tenure and address any drivers performing below the benchmark. Statistically, these segments have a higher risk of attrition, despite also having higher levels of engagement relative to the rest of the tenure segments. 

6. Improving the prediction accuracy

Peakon’s Attrition Risk model works most accurately when the following is implemented:

  1. Increasing active survey participation whilst also actioning feedback 
  2. Moving to a higher survey frequency to identify trends and risks earlier - the more recent the answers, the more accurate the prediction of current risk. 
  3. Enabling Peakon’s standard ‘Loyalty’ question
  4. Collecting leaver data using Peakon’s Separation date and Separation reason attributes

The above points are not required for the Attrition Risk feature to work. However, enabling them will improve your score accuracy and contribute to the overall anonymized data set used in the model.

7. Enabling Attrition Prediction

Attrition prediction is enabled by default for all administrators (the Administrator Access control group). To enable this for other users on Peakon:

  1. Click on “Administration” in the left menu
  2. Select “Access control” 
  3. Select the Access control group you’d like to enable it for
  4. Under the “Access statistics” permissions toggle on the “Attrition prediction” permission

Article: Using the Separation date and Separation reason attributes
Article: Employee experience cycle
Article: Questions library and theory references

Did this answer your question?