Skip to main content
Gainsight Inc.

Impact Analyzer FAQs

 

Gainsight NXT

This article supports Gainsight NXT, the next evolution of the Customer Success platform. If you are using Gainsight CS Salesforce Edition, you can find supporting documentation by visiting the home page, and selecting CS > Salesforce Edition.

Not sure what your team is using? Click here.

 

What are the advantages of Impact Analyzer?

Optimize your Customer Success strategy by uncovering the impact on top business goals and identifying where your team should focus their efforts with Impact Analyzer.

Is this available in both Salesforce and NXT?

Yes, it will be available in both editions.

How do I enable Impact Analyzer on my existing Gainsight subscription?

Impact Analyzer is available in the CX Center and will be auto-enabled. Admins are not required to perform any prerequisite actions to enable the feature.

Key Drivers help identify the underlying factors that affect the key customer experience metrics (called outcomes) such as NPS® and CSAT, by classifying them as strengths and opportunities (weaknesses). 

Underlying factors, called drivers, are derived from:

  • Company attributes (like size, ARR, number of employees etc)
  • Scorecard measures
  • Survey responses 

Note: Historical data from Scorecards and Survey responses (from the last six months) are considered for the analysis, making it thorough and comprehensive.

Additional drivers such as person attributes and CSM activities will be added in the future.

What sources of information are being pulled to create the Impact Analysis?

Data sources for Impact Analyzer are Survey responses, Scorecards, and company attributes stored in Gainsight's Company object - both standard fields and custom fields.

All potential drivers are categorized as strengths and opportunities.

This distinction is made based on the importance and performance scores. Both Strengths and Areas of improvement are important, and help you to identify potential areas you can focus on. Areas of improvement have more scope for refinement compared to the strengths and might be the first place to look at, if you are looking to improve the outcome.

  • Importance: To figure out the importance, we try to identify whether or not a relationship that exists between a potential driver and an outcome is strong. If a strong relationship is found, it becomes important.

  • Performance: Performance refers to the score based on  the relationships between the outcome and the driver. This also refers to what percentage of the driver falls under areas that have high outcome values; the Performance score increases with the higher percentage. For example, if a green health score means a high NPS, performance looks at what percentage of responses are from companies with green health score.

Importance: A second degree curve has been fit between the outcome and driver. This ensures that the data is strongly adhered to the curve, subject to the following constraints:

  • Scorecard variables: A monotonically increasing curve is fit. As the scorecard measure value increases, the outcome value should also increase

  • Survey Questions: Increasing and decreasing curves are only fit. U-shaped curves and inverted-U shaped curves are not used

  • Company Attributes: No restrictions

Performance: The performance score is the correlation between the outcome-driver relationship curve and the distribution of the driver. The higher the value, the more likely it is that the driver is distributed in areas where the outcome’s value is high.

Predictive Power indicates what portion of your data is capable of explaining certain patterns. It’s important to note that a low predictive power does not mean that the analysis is unreliable as such. Gainsight is always working to strengthen our logic making predictive power’s buckets (low/high) more meaningful. Predictive power refers to the overall quality of the data underlying the analysis. 

  • A high-predictive power means that the drivers considered explain a large portion of the variance in the outcome, and makes the overall analysis highly reliable. 

  • A low-predictive power indicates that the drivers considered explain only a small portion of the outcome, and deems the overall analysis less reliable

Predictive Power is the R-squared value which is received after fitting a model on the outcome using all drivers.

 The following two options can be done to improve the predictive power:

  • More data points: bring in more survey responses if possible

  • More attributes in the Company object: if other attributes are deemed important, the attributes can be added as company attributes, and then included in Key Drivers.

Fields of numeric and date data type attributes are supported in Key Drivers today. The analysis by default covers ARR, Number of Employees and Customer Lifetime in Months. You can change this configuration through the Administration page.

Yes, all survey data loaded into Gainsight’s survey data model.

Key terminologies used in Impact Analyzer > Key Drivers:

  • Outcomes: Output of Impact Analyzer’s Key Drivers  (NPS and CSAT) are called outcomes

  • Drivers: Input to Impact Analyzer are called drivers

Yes, the driver can be excluded from the analysis by leveraging the toggle next to the driver on the View Full Analysis page. This action is not supported on the default view.

clipboard_e7c073baf2f86bea2050c2a0841bd2c9d.png

The summary ribbon on top of the View Full Analysis page shows the drivers excluded by our model. Clicking on it will reveal the full list of excluded drivers along with the reason for exclusion.

clipboard_e2b84327616c9dd84ebd531ce2d25307a.png

Yes.

How many responses do we need for the analysis?

The model requires at least six responses for the analysis to run.

These drivers are excluded by the model due to Poor quality or insufficiency of data. 

No. You cannot change the way the importance and performance scores are computed because they are calculated by the model.

Predictive Power is an indication of the overall confidence of the model based on the quality of data supplied. Providing better quality data might help improve Predictive Power of the analysis.

The model updates all analyses once a day. Any admin configuration changes, however, are updated real-time in the product.

These drivers share a strong relationship, and improving one driver improves the others.

How can I feel confident building strategies and recommendations based on the data I receive from Impact Analyzer?

Gainsight has taken extra care to remove accidental correlations to ensure that the correlations presented to you are accurate and impactful.

 

  • Was this article helpful?