DS-301d · Module 3

Data Quality for KPIs

3 min read

A KPI built on unreliable data produces unreliable decisions with mathematical precision. Data quality for KPIs requires three guarantees. Completeness: every relevant data point is captured. If 15% of deals are not logged in the CRM, the pipeline metric is 15% wrong. Accuracy: the captured data is correct. If reps advance deals to the wrong stage, stage-based metrics are miscalibrated. Timeliness: the data is current. If CRM updates lag by a week, the real-time dashboard is showing last week's reality. Each guarantee requires its own monitoring mechanism. Completeness is monitored by comparing record counts against expected volumes. Accuracy is monitored by sampling and verification. Timeliness is monitored by tracking the lag between event occurrence and data appearance.

Do This

  • Monitor data quality with the same rigor as the KPIs themselves — bad data produces bad metrics
  • Publish data quality scores alongside KPIs so decision-makers know the reliability of the numbers
  • Automate data quality checks — manual verification does not scale

Avoid This

  • Assume the data is correct because it is in the system — systems contain whatever was entered
  • Publish KPIs without data quality context — a metric with 80% data completeness should be treated differently than one with 99%
  • Fix data quality issues only when someone notices an error — systematic monitoring catches errors before they reach the dashboard