
332 Part II n The Step-by-Step Guide to the TIQM Quality System
Retrieve
(error)
Re-
“create”
Interface
Correction,
Interface
Ongoing
Data
Correction
Uncorrectable
Error
Errors
produced
Defective
Actions
Defective
Actions /
Decisions
Propagated
Errors
Defective
Actions
Strategic
Database
Defective
Create /
Update
Process
Retrieve
(error)
Re-create
(error)
Retrieve
(error)
Decision
making
4
3
2
5
6
1
Better
Decisions
Figure 8-3: Sub-optimized Data Correction for the Data Warehouse Only
This scenario for data correction is as follows:
1. Broken processes create errors at the source, but data correction at the
source of data is rejected as out-of-scope. The false thinking is that “we
only need to correct data as it comes into the data warehouse.”
2. Errors are propagated from Record-of-Origin Database to Record-of-
Reference Database that feeds the data warehouse.
3. Massive, costly data correction activity is implemented, using software
that looks at reference databases to correct discovered errors, often intro-
ducing errors and often failing to identify errors.
4. Corrected data is propagated to the data warehouse along with undis-
covered and uncorrectable errors.
5. Uncorrectable errors will be propagated to the data warehouse without
being able to be corrected.
6. Better — not necessarily optimal — decisions are able to be made, but
only for Knowledge Workers using the data warehouse.
Note Not all errors may ...