The problem of data quality billion; Big & # 39: 3 tips to avoid this

Image: iStock

It is estimated that the poor quality data costs businesses $ 600 billion per year. It is not just the potential for serious mistakes that bad data leads, but also the amount of time and labor-intensive human effort it takes to correct this data ,

In the large data world, the data can be multiplied exponentially data quality problems. Web data is generated, known for its problems of lack of reliability and quality, and then paste the data characterized by the machine that the Internet of Things (IoT) generated; Machine data generated may contain gibberish as useless as the status information is priceless.

The problem was data highlighted by the consultant Thomas C. Redmon in an article in the Harvard Business Review in 2013 Redmon uses the example of a framework for key products to support to a report of its senior team to prepare the office and noted that the figures market share the report did not make sense. The Board asked an assistant to check the numbers, and the wizard has detected an error in the data that is found provided subsequently corrected by the market research department. The good news is that the wizard called the mistake in time. The not so good news is cited studies in the same article revealed that lost "working knowledge of up to 50 percent of their time on the hunt for information, identification and correction of errors, and sources of research confirms data they do not trust. "

There are tools for data preparation Summary transform and extract formula loading (ETL) of great cleansing of data to end users, including the company can use, but there are still some problems of data quality, while creating the data occur. It can be as simple as someone from the business transactions hastily drilling simply by entering into the system, not knowing that someone in the Business Unit B, which is at the other end will have this flood of data to clean and needed to the analysis and implement data to make better business decisions.

Or it could be a machine spits dice random bits and bytes, how it operates, so that the user the process of removing data containers, exit to isolate the elements Useful information.

This mixture also "Champions own data" includes the cost of working with dirty data and then try to fix it. But anyone who has served in the own data master knows the task is misunderstood functions and that few executives and / or corporate reward systems to recognize itself.

There are three steps companies can take to improve the quality of their data.

Key data streams Link business processes: 1

If the data and commands via the web on the entry system internal organization with marketing analysis is produced, has marketing are closely related to data sources (for example, the input group command and supplier of data generated on the network), so that everyone is working in a pipeline of data which is started the computer when the data is generated, and ends when it is consumed. Thus, the problems of data quality can be detected and corrected from end to end, before they become overwhelming responds.

2: data standards and guidelines for the use of their data Preparation Tools

More and more companies adopting data preparation tools, self-service departments to work with their data quality without consulting IT end users; the problem is that inconsistencies occur in the practice of data. A standard method for the proper use of these tools must be defined by IT, the historical experience with data cleaning and preparation has.

3: Move more responsibility for specific end business data

IT is the custodian of most of the data of the company but not its creator, nor its main users. If the quality of the transaction data and great prosperity are executives finally tired to be violated for the first time in the company of the daily errors and loss of time in trying to the situation, which is consumed to correct the data.

Also take a look

This entry passed through the Full-Text RSS service - if this is to read your content and on the website of another person, please read the FAQ on fivefilters.org/content-only/faq .php # publishers.