¿Cómo se mantiene una calidad de datos coherente y escalable?

There are a few more things that should occur prior to defining your data quality criteria. Such as: 1. Analyze your information environment, including the various systems and their dependencies, but also make sure you have an understanding of the business and cultural one. These will provide you with the right leads for determining the root cause of bad data quality. 2. Analyze the business impact. Do an initial measurement so you can better determine the impact your data quality would have on the business if left unresolved. It can be tricky to determine the cost of missed opportunities, but you can determine cost savings for things like reducing product defects or increasing efficiencies in data flow processes.

It stands to consider aligning data quality criteria with the key performance indicators (KPIs) that matter most to the organization, ensuring that data accuracy, completeness, consistency, and timeliness directly support the measurement of these KPIs. Further, consider involving representatives from all parts of the organization when defining these criteria. The insights from different teams can provide diverse perspectives on data quality needs, leading to a more comprehensive and practical set of standards. Lastly, it's crucial to ensure these criteria are adaptable and scalable. What is essential for the business today could change in the future, so the data quality criteria should integrate those changes.

It's also important to determine for which data sets and attributes you are going to measure quality. It's not possible, nor sensible to attempt to measure data quality for all of your data, so a prioritisation exercise based upon business impact needs to take place.

Identifying the root cause of bad data quality is one of the most important things to undertake as part of improving your data quality. Otherwise the issues will keep on occurring. There are several approaches for root cause analysis one could take here: -Safety-based -Production-based -Process-based -Failure-based -Systems-based Or a combination of the above. You can apply different techniques, such as the fishbone diagram exercise or the 5 whys.

In my experience, this step is very crucial. We have implemented many data-quality dashboards and controls, which are used to check the health of the data on a weekly basis by data stewards. Also monthly checkups at the data governance council level. This helps to identify any other new issues and also gives confidence to the business.

AI and machine learning could enhance the process of assessing data quality. Automated Error Detection: Machine learning algorithms can be trained to identify common errors or anomalies in datasets, thereby streamlining the process of finding data quality issues. They can recognize patterns that might be overlooked in a manual review, allowing for a more thorough and accurate assessment. Predictive Data Quality: Machine learning can also predict future data quality issues based on historical patterns. This predictive capability can enable proactive measures to prevent those issues.

Modern tools can offer vast help in maintaining high data quality standards. Data Governance Tools, for example, help enforce data governance policies across the organization. Data Validation Tools can check for anomalies in the data by applying preset rules and logic. Data Lineage Tools provide an overview of where the data comes from, how it's processed, and where it's used. On the other hand, consistent and scalable data quality doesn't only depend on technology but also on the culture of the organization, which include continuous education, a clear understanding of roles and responsibilities, and an emphasis on the importance of high-quality data in every aspect of the business.

Preventing bad data quality from occurring or re-occurring can be a hefty step as it would include: -Identifying and solving the root cause of the issue -Redefining the business process -Adding data validation rules at point of entry -Adding or rewriting your ETL(s) -Prepare training materials and sessions in place to prevent bad data from reappearing -Implement other change management techniques Last, but not least, certain data quality audits also need to be established to monitor your data quality status and identify those exceptions when need be. Luckily, most of the scripts created for assessing the data quality level can be re-purposed and transformed into audit reports.

Prevention is better than Cure, so we must identify the root causes and fix them permanently as mentioned below. Most of the time it is either system issues or training/process issues, both can be easy handle.

You need to understand your current state. There are several data quality tools on the market that will scan, profile, and score your DQ. Informatica’s Data Management Platform has this technology available out of the box. This will also help you determine which data assets need the most attention based on usage and quality.

Along with technology and regular assessment, cultivating a data quality mindset within the organization is fundamental. It requires a shift in attitudes where everyone views high-quality data as a crucial part of their role. For example, regular training programs can be organized to educate employees about the importance of data quality. It is also important to designate data stewards to help ensure responsibility and accountability for data quality. Each team member should understand their role in maintaining data quality. And don't forget to recognize and reward teams or individuals who contribute significantly to maintaining and improving data quality.

It is sine qua non to understand data , secondly draw the standards and involve all stake holders . Put the process , set the standards . Try automation in phases Make a step which checks some vital parameters like date of which the data belongs , the records counts Do these preliminary checking before the load , we can term the process as pre load audit checks . Similarly do post load audit checks Make as part of pipe line. Lastly set standards for vital data elements and run checks and collect the data and then visualize it .

Whilst roles and responsibilities are important, try to make Data Quality something that the business strives towards in a collective sense. It's still too common to see a centralised data team coming under constant criticism for an unrealistic expectation that they will somehow ensure every piece of data in a firm is accurate.