Get a first look at what's coming. The Developer Passport Australia Release Preview kicks off March 12. Dive in! 

How is the “Suggested” target value calculated in Proactive Analytics?

Kohei Tominaga1
Kilo Sage

Hello ServiceNow Community,

 

In Proactive Analytics target configuration, the Value field is pre‑populated as “Suggested”, as described in the documentation below:
https://www.servicenow.com/docs/r/now-intelligence/proactive-analytics/proactive-analytics.html

 

However, I could not find details explaining how this suggested value is calculated.

Could you please clarify:

  • What logic or statistical method is used to derive the “Suggested” value?
  • Is it based on historical data (average, trend, percentile, etc.)?
  • What time range of data is evaluated?
  • Does this value adjust over time or is it fixed at creation?

Understanding this would help administrators decide whether to use the suggested value and explain the rationale when questions arise.

 

Thank you for your support.

4 REPLIES 4

Tanushree Maiti
Tera Sage

Hi @Kohei Tominaga1 ,

From your shared documentation, it seems it is using Predictive Intelligence at backend .

 

ServiceNow uses Predictive Intelligence (PI) to analyze historical data and suggest values for fields (e.g., Assignment Group, Category) or suggest solutions (e.g., Knowledge Articles).
 
  • Machine Learning Models: PI uses classification and similarity algorithms trained on historical data, such as closed incidents.
  • Precision and Recall: The model's effectiveness is calculated based on:
    • Precision: How many of the "Proposed" (suggested) values were actually correct (reducing False Positives).
    • Recall: How many of the total records that should have been classified were correctly identified (reducing False Negatives/misses).
  • Confidence Thresholds: The system only suggests a value if the probability of accuracy exceeds a predefined threshold

 

Please mark this response as Helpful & Accept it as solution if it assisted you with your question.
Regards
Tanushree Maiti
ServiceNow Technical Architect
Linkedin:

Hi, @Tanushree Maiti 

Thank you for the explanation — I agree that Predictive Intelligence and machine learning techniques are likely used in the background, and your description matches my understanding.

What I am trying to clarify further is how the “Suggested” value is calculated specifically as a target.

If the suggested value is simply a reflection of historical performance, that makes sense.
However, if there is logic that proposes an improved or aspirational target derived from past trends and model outputs, I would like to better understand the underlying approach or formula used to arrive at that value.

For example:

  • Is the target adjusted upward or downward based on historical trends?
  • Does it incorporate optimization logic beyond a simple prediction?

This level of detail would help administrators confidently explain the rationale behind the suggested target and decide when it should be adjusted.

Thanks again for sharing your insights.

Hi @Kohei Tominaga1 ,

I guess you are looking for how to finetune the predictive intelligence models. Though I have never worked on this deeper level, but following Community articles can help help you.

 

https://www.servicenow.com/community/intelligence-ml-articles/tuning-predictive-intelligence-models-...

 

https://www.servicenow.com/community/intelligence-ml-articles/tuning-predictive-intelligence-models-...

 

https://www.servicenow.com/community/intelligence-ml-articles/tuning-predictive-intelligence-models-...

 

https://www.servicenow.com/community/intelligence-ml-articles/tuning-predictive-intelligence-models-...

 

https://www.servicenow.com/community/platform-analytics-articles/tuning-predictive-intelligence-mode...

Please mark this response as Helpful & Accept it as solution if it assisted you with your question.
Regards
Tanushree Maiti
ServiceNow Technical Architect
Linkedin:

No, I don't want to make any change but want to know the details how OOTB calculate suggested value...

I have searched a lot of sites but could not find any resources