Improvement of Data Quality = Better Business Intelligence

Data Quality - The way to a better quality ofgeneral. There are several steps that need taking
Business Intelligencecare and who should be found in every project to
Relatively little attention during the various BIimprove the quality of data...
(Business intelligence) projects is payed on data1.Setting Team and Resources
quality from production systems. This is data uponIn order to effectively cope with the problems of
business decisions will be made. Sourcedata quality and access problems company needs
(production) systems are basis of information andto form a team of IT staff and staff on the
are feeding BI applications and with aggregationsource data, and reports of people who are
and presentation of data in a certain way. If theexpecting, ie, lower management. The technology
input data does not meet certain quality levels it iscan be helpful but it is necessary to adhere
unrealistic to expect that the usefulness ofmanual work, which typically consume valuable
projects and applications that have occurred onresources in the company, people who need to
such a precarious be well, even though their ownreview and based on his experience and feeling
projects and applications can be done technicallysay that such and such data, these certainly are
perfect. Regardless of whether the datanot good, that historical data should be
warehouse project, planning, or a project thatsupplemented with some elements and so IT
provides a unique overview of the service users,staff can in principle be of assistance in terms of
the quality of existing data will most directlyautomation manual entries, etc. but we should not
affect the result of the project. Poor data qualityexpect that someone will do some magic
at source will surely cause poor business decisions.program that fixes incorrectly entered data.
Once it is proven how available data is of2. Establishment of Quality Metrics
low-quality, applications users would often leaveFocusing on the overall quality of the data in the
the project instead of dealing with improving theenterprise is of course non effective. Some areas
quality and to recognize that a key problem in theare certainly important in making decisions and
functioning and success of the BI, and similarsome are less important. Some IT segments
projects. Regardless of the integration projectsdealing with the basic activity of enterprises is
poor data quality has its influence also in theusually with less errors, such as billing, which is
production systems. Consequences are are usuallysubject to customer complaints and is easy to
manifested as poorer productivity with morespot errors. Should focus primarily on those key
errors when routine tasks that use the data ofareas and provide them with appropriate priorities
operating system (easiest example: billing to thein the short term will most contribute to the
wrong address). So production shows inability toquality. It can be filling half empty information
provide information for monitoring businessfrom a variety of reasons usually not included,
activities generally and/or those jobs require andrecast some data that are often known to be as
consume a lot of IT resources in terms of humanthey should, or the introduction of completely new
labor.attributes that changed the character of the job
The fact is that few companies have anrequires. Focusing contributions and quality metrics,
awareness of how the data is poor quality andie the system by which we measure the quality
much less any awareness that something needsof data. It can be a very simple test where you
to be done and that the quality of the data mustcalculate how much the availability attributes x in
be treated as an equal business problem. Mostthe table customers through simple tests to date
businesses could specify at least one of thethe cancellation fee is not greater than the date
project, which is inefficient in the sense that it isof entry to the much more complex tools that
not used because the data with which to operatehelp detecting entry errors, etc. Identification of
there are not completely accurate, ie can not bethe major problems, which later quantified (eg
placed upon them. Improving this situation is the30% of the stores where there are no specific
explanation that the problem lies in the quality ofindustry or belong to more than 70% of
data and not in the Business Intelligence project.members there is no information entered is used
Incomplete data and low qualityto the Internet and / or electronic mail) is a good
The most common problems can be observed asbeginning and a clear goal of what to do as
incomplete or poor quality data. It means that theopposed to general statements that should be
data simply does not exist or another that data iscompleted for all non-existent information the
inaccurate and that is worse. Incorrect data is ofmember / customer.
course dangerous because it seems everything is3. Divide complex tasks into elementary
OK and actually bringing it as basis for a wrongIt is easy to focus on small elements of the
decision. BI applications are mostly dealing withproblem and gradually solve them everywhere.
agregated results and presentation of such results.Previously described examples show that it is
Incorrect data for a large degree of aggregationgood to focus on several attributes of a table
will go unnoticed more often than in the operatingwhich are known to be important in
system.decision-making process and that it is possible to
For example, if user X has an error to accountamend / correct the information we have. It is
for 10.220 USD instead of 1022 USD to bequite another problem if the information is not
produced will be probably noticed in productiongenerally available (or are available but they need
system because the number of accounts in suchto buy) or if not available, eg Customer is gone
an account 'stick' among others and ultimatelyand company no longer know any information
appeal administrator. After millions of aggregatedabout him. Company can develop good way to
amount in the BI application, the difference ofgather information on point of sale where
tens of thousand is not clearly visible.customers could voluntarily provide many useful
Incomplete data is an illusion that we have theinformation (for example software industry where
information. This is a dangerous illusion inthe registration of products and benefits brings to
applications where the design should be consideredan end user a good reason for some feedback
to have some information and when it is time toabout themselves and their vision of the product).
analyze and present (which is always at a later''Training and''cultivating staff to enter information
stage of the BI) turns out to be no. For example,and training staff to recognize that some
management company thinks it has the e-mailinformation, although not mandatory does not
addresses of its users, in fact, have about 3% ofmean that they should not be entered through a
e-mail users because the application does notperiod, and so to get better data. The staff
require this field to always be filled (which in thisengaged in data entry are often neglected fields in
case is the only correct approach to thisthe optional programs that are not optional entry
attribute). With such data can not learn somethingso that they can not enter, but primarily because
new about the structure of the user and can beit does not attribute any such its value in the real
used in further analysis for sales and marketing.world.
Data Warehouse4. Measurement of results
Data Warehouse collects data from productionTo be able to tell how well the job done using the
sources in the manner already described in otherpreviously defined metrics, we now have a tool
articles of the author. The analogy with ordinarythat describes quality entry clerks and the degree
warehouse is applicable in case of incomplete fillingof realization of the project. Most of the time
of data warehouses and / or inaccurate data.course carries project time officers and other
Data Warehouse filled with incomplete data, it isstaff have expended to retrieve data.
very easy to fill, ie additions. It is necessary "onConclusion - data quality and business intelligence
the shelf" only add what is missing, clearly theIt is usual that in the eyes of senior management
case that this can be obtained from somewhere.IT is blamed for the quality of reports failed and
Data Warehouse filled with wrong information isdid not prove expensive investment of BI. Then it
like a warehouse with mixed items where ahappens that forcing IT management to clean
worker when it comes to some shelves theredirty data. It is a task that goes beyond the role
are items that do not belong there and haveof IT that should allow the flow of information,
nothing to do with the description on the shelf.storage and access. According to Gartner, about
This warehouse is much less usable than the80% of the process to improve the quality of
empty and dangerous. It requires thendata guided by IT will be ineffective in achieving
considerably more time to find what you needgoals. IT does not invent, and design information
and decide whether this is actually what we want,as well as trade information within the company
because in such a repository has no order.and shall update and amend data. Senior
Steps to improve data quality in businessmanagement must be aware of strategic
intelligenceimportance to the quality of input data and it is
Once the responsible people in the company tonot just a problem for the overall IT operations
recognize that efforts to improve the quality ofand the entire company. Also it is a process that
data is not foolishly spend money, but the road tomust always be continous and not be a project
better quality information in an enterprise infor only a specific period of time.