Delivery operations data quality: the foundation behind reliable performance reporting.

Driver identity is a data-quality problem

The same person can appear differently across operational systems: full name, shortened name, transporter ID, encrypted identifier or another source-specific reference. If those identities are not reconciled, the same driver can be split into several records or matched to the wrong history.

Identity mapping should therefore be explicit, reusable and auditable rather than hidden inside one import.

Duplicate data should be detected, not silently averaged

Duplicate rows can be legitimate when a source contains multiple events for one driver, but they can also result from repeated uploads or inconsistent source files.

The system should preserve enough source information to explain why a duplicate exists before it affects a driver or site metric.

Missing is not the same as zero

A missing value, a zero score and a report that does not contain the metric are three different states. Treating them as equivalent can make a dashboard look complete while changing the underlying meaning.

Reliable analytics should retain the distinction and surface incomplete evidence where it matters.

Keep reporting periods consistent

Daily and weekly data should not be mixed without an explicit rule. The same metric can mean something different when it represents one day, a rolling period or an official weekly scorecard.

Store the reporting date, reporting granularity and source file alongside the metric so later analysis can compare like with like.

Make data-quality issues visible to managers

• Show unmatched driver records that require reconciliation.
• Flag missing or stale source data before the dashboard is interpreted.
• Keep source-file and reporting-period context available for investigation.
• Track manual mappings so repeated imports become more reliable over time.