Data Quality Management
Why checking data's quality?
In the short term, implementing data quality management has a cost. However, in the long term these costs are largely offset by operational gains and new business opportunities.
Internal requirements
-
avoid making bad decisions
-
improve operational performance
-
be more competitive by quickly reorienting the strategy
External requirements
-
satisfy customers
-
comply with regulations
What makes quality data?
To evaluate the quality of a data, it is necessary to define it's purpose. However, for any data, a set of intrinsic characteristics must always be checked.
Up-to-date
Availability
Functional & technical
consistency
Traceability
Comprehensiveness
Security
To ensure the functional and technical consistency of the data, it is necessary to verify:
-
Validity of data format
-
Cleaning of non qualitative data
-
Homogeneity of data
-
Deduplication of data
-
Data enhancement
How to get quality data?
1
Manage with an integrated platform
Transactionnal database to ensure homogeneity and deduplication
Generalized audit and integrated documentation for full transparency
2
Define the scope of the controls
Completeness and consistency check
Standard or customizable validation
Global or sampled control
Low or high alert thresholds
3
Validate at each stage of the process
When importing or encoding data
When changing or cleaning data
Before and after calculating analytics
When sharing or publishing reports
4
Automate controls regularly
Process scheduling by portfolio or by business
Frequency definition by complexity or by SLA
Error-based automation or 4 eyes validation
E-mail alert and user-friendly dashboards
Dashboard
5
Set up and monitor KPIs
Graphical display of executions
pre-and post-production
Follow KPI by portfolio or by process
Improve planning and control integration