Databases Reference
In-Depth Information
the increased value of the system to its users might well conclude the 5 per-
cent error rate is acceptable.
On the other hand, what if that error rate was actually 30 percent? Or
what if the application in question was sending bills to your customers? It
is unlikely that faulty data could be tolerated for long in either situation.
Just because existing systems can use a particular set of data, that does
not necessarily imply ease-of-use for a new system — among other things,
there may be a great deal of hard-coded logic in the old systems that allows
them to bypass problems in the raw data. And there are a host of potential
new problems once you start merging data from several different existing
systems in order to feed your new system.
And postponing the cleansing effort is generally inadvisable. The initial
data propagation/transformation efforts can require significant revision
and re-execution. Considerable additional effort and cost may be needed to
repair data, to make it accessible, and to prevent continued corruption.
However, once the users have started to access the system, its credibility
(and your own) are on the line. There may not be any funds available for
continued work if the system's users are not satisfied with the initial re-
lease. And does time and additional resources ever
really
become easier to
get onto a project at a “later date”?
Potential Impacts
Some potential impacts of unreliable data include the following:
•
Increased Cost of Database Marketing.
Poor quality data affects di-
rect-mail, telemarketing, and marketing collateral costs. A large auto-
mobile manufacturer purged its customer files and cut direct-mail
costs by U.S. $500,000 the first year, and lowered the costs of managing
this data long term.
1
•
The time that resources must spend to
detect and correct errors can be very significant. One estimate for the
cost of poor data quality is 8 to 12 percent of revenue; for service or-
ganizations, a “good working estimate” might be 40 to 60 percent of ex-
penses.
Increased Operational Costs.
2
There is also a related negative impact on employee morale.
•
Multiple mailings sent to the same
customer can be annoying, they can cause the customer to question
your enterprise's competence, and can harm the customer relation-
ship. This is especially true if the information within a given mailing or
statement pertaining to a customer is incorrect.
Lowered Customer Satisfaction.
•
Reduced Use of System.
If the believability and/or the objectivity of
the data comes into question, then the resulting poor reputation will
cause reduced use of the system and underlying databases.
Consider-
ing the effort and cost of developing these systems, anything impair-
ing their value or use is clearly counterproductive.
3
Search WWH ::
Custom Search