Confidence Score in SmartResearch

Purpose

This article explains the concept of the confidence score in SmartResearch. It describes what the confidence score is, how it is calculated, where it is displayed, and how users should interpret and act on it.


What Is a Confidence Score?

A confidence score is a numerical or qualitative indicator that reflects the reliability and trustworthiness of a metric or analytical result in SmartResearch. The confidence score helps users quickly assess whether a metric can be used for decision-making or if further review is needed.


How Is the Confidence Score Calculated?

The confidence score is automatically generated by SmartResearch’s data quality engine. It is based on several factors, including:

  • Data Freshness:
    How recent and up-to-date the underlying data is.

  • Source Consistency:
    The degree of agreement between multiple connected data sources (e.g., ERP, external databases).

  • Data Completeness:
    The presence or absence of missing, incomplete, or anomalous values in the data.

  • Data Integrity Checks:
    Results of automated validation and reconciliation processes.

The confidence score may be displayed as a percentage (e.g., 95%), a qualitative label (e.g., High, Medium, Low), or both, depending on your organization’s configuration.


Where Is the Confidence Score Displayed?

  • The confidence score appears alongside each metric value in the Metrics or Analytics section of SmartResearch.

  • In the metric detail view, the confidence score is shown near the top of the panel, often with a tooltip or link to more information.


How Should Users Interpret the Confidence Score?

  • High Confidence (e.g., 90–100% or “High”):
    The metric is reliable and can be used for business decisions.

  • Medium Confidence (e.g., 70–89% or “Medium”):
    The metric is generally reliable, but users should review the underlying data or consult with a data steward if the metric is critical.

  • Low Confidence (e.g., below 70% or “Low”):
    The metric may be based on incomplete or inconsistent data. Investigate data sources, review data quality warnings, or contact your SmartResearch administrator.