|
NOTE- You may refer Data Quality Policy in Execution-MiH Data Quality Practice Tool-Kit for more details. In-brief, we have stated that Data Quality policy is the first step to drive your data quality agenda. The question here is that do we need to address all the headings as mentioned in our data quality policy template. The context could be the organization may not be ready to implement or follow some points mentioned in the data quality policy. To rephrase the question- if we know that we cannot do something (given the readiness of the organization), does it make sense to put into data quality policy, and thereby create a credibility issue?
Yes. You need to have all components of the data quality policy as mentioned in our template. With any component missing, your foundation for Data Quality practice will loose its strengths and will therefore not be sustainable. Here are the reasons:
Business Owner Ship of Data Quality:
Business needs to be the owner as business is the reason as well as the impacted party for the data quality issues. Here are some examples:
- More than 80% of the specifications related to the data quality assurance come from business (like data models, data standards, input form validations, business rules related to data etc...).
- Data quality is not only limited to IT systems, but also manual business processes. Therefore, the quality of these manual processes also decides upon the health of data.
- Business is the ultimate beneficiary (or sufferer) of state of data quality.
- Business is the one, which sponsors the funds for Data Quality program.
- There are many informal systems running within the business units, which are not maintained by IT, and they are typically having loose DQ controls. These systems not only are a risk to themselves, but also may feed faulty data into the core systems.
In-short, if business do not own the data quality, it will be against the natural law of stake holding.
Business Custodian-ship for Data Quality
If we ask a question- Who should be the custodian of customer credit card information, irrespective of its form and location? The answer cannot be IT. This is because any data can be in manual form, in excel sheets, in informal applications and core IT systems. Therefore it has to be business, who has to be the custodian. While, IT may be doing it, it’s the data custodian who has to be responsible for archiving, purging of data. Custodian has to decide on the data which is going to be online vs. offline. Data Custodian is also responsible for the disaster recovery for the data. In other words, if some data goes missing or is not available as per the need, data custodian should own it up.
Appointment of a Data Steward
If you consider Data Quality as an enterprise level domain, one has to have an enterprise level role for the same. If you do not have an enterprise level role and data quality responsibility is scattered, it will be impossible to drive enterprise level initiatives like DQ assurance guidelines. When we go to our client organization on a data quality initiative, this is the first step we ask them to implement. (Refer Data Steward)
Data Management Council
Data Quality should not be a management team subject, unless there is a question of funding or some major crises. Given that a management team has many other priorities, the subject of data quality has a risk of being ignored in management meetings. One needs to have a data management council which is focused on the overall data management (master data management, metadata management, data integration and data quality). This council typically comprises business owners of data-groups (refer Data-Group Master in our Data Quality tool-kit package). If Data Management council does not exist, Data Steward will not have any audience of reasonable gravitas. Data Council will decide on funding and priorities of Data Management initiatives.
Data Quality Assurance and Control guidelines
This is a no brainer and can be achieved quickly. These are the guidelines and are a basic hygienic component of a data quality policy. (Refer DQ assurance and control guidelines tool)
Data Management Universal Standards
This is what takes organizations most of the time. In-brief, an organization needs to create the data management standards (domain value standards, data format standards, data model standards and business rules...) for data entities (like customer entity, invoice entity, product entity...). This ensures a consistency on how we process, store and interpret the data in our enterprise. Many of our clients have asked us on if it is needed to be placed as part of the policy, as it will take a long time to develop a thing like this.
(For more details-Refer Data Management standards for data entities in our tool-kit)
Data quality Assurance Procedures
This is nothing big and can be done fairly quickly and evolved. This includes procedures related to ensuring data quality in an initiative, data monitoring procedures, data correction procedures etc. The point to note is that in typical organization, most of these procedures will exist. For example, data back-up procedures, data migration procedures, release management procedures etc...
Data Audit and Monitoring
This is again nothing out of the way, which organizations don't do. Most of the organizations have data audit and monitoring done on regular basis. The area where we have seen organizations falling short is on how they do it. However, the quality of this exercise has not much to do with its mention in the data quality policy. Therefore it should stay.
NOTE- We have provided some tools related to data mapping and assessment and overall system health assessment, to facilitate.
Data Correction
This is must to place, as we believe that a good proportion of data quality issue happens when organizations inadvertently create a bigger issue to clean-up a small one. Data correction from the back-end or front-end has to be carefully monitored.
Data Quality Measurement
If you don't measure something, it will not get done. If data quality is not in the goal-sheets of key individuals, it will not get that importance. Yes, you may vary the frequency and scope of that measurement, given the level of organization readiness.
Data Quality Awareness
This is a fairly non-controversial point as it is to do with the necessity of bringing awareness on data quality across all levels of the organization.
OVERALL- our strong opinion will be that leaving any of the above points out of your data quality policy (you may rather add more) will leave a visible gap in your data quality agenda.
|
