Data quality is objectively a good thing—I doubt there’s a business that doesn’t agree with that fact. But how do you know what quality data looks like? How can you assess your data quality to determine how your data stacks up?
Without data quality testing, you won’t know what potential data issues you may have, or how significant they are, making it extremely difficult to work toward the right solutions. That’s why it is essential to create a data quality checklist for your business before attempting to conduct a data quality audit.
Data quality criteria
There isn't a single, definied list of data quality checks that exists—it’s very business- and function-specific. Defining what you do with your data and what you need it to do truly informs your evaluation.
That being said, there are some common areas to consider when thinking about what to include on your data quality checklist:
Data quality testing
Once you’ve determined what broad criteria matters to your business, you can plan for a data quality test. There are several steps in this process.
1. Define specific data quality metrics
It's not enough to have broad data quality criteria—you need specific metrics to test against. Consider the following: What type of data is it? What are you doing with it? Think through your business purpose for the data to help define specific metrics that impact your business operations. Some examples include:
Data quality metrics that matter will vary based on your job role or focus area. If you are an email marketer, your gauge for data quality may be how many email addresses on your list are reachable. If you're part of the call center, you probably care more about collecting valid phone numbers than email addresses.
2. Conduct a test to find your baseline
Without defining your baseline state, you can’t really drive forward data quality improvement. In our example of email addresses, there are specific tools available to help you easily assess your email list data quality. The result is a report that defines how bad your problem actually is.
In our example, we’ve found that there are potential issues with about 35 percent of emails on this list—quite a significant percentage that needs to be addressed.
3. Try a solution
Once you’ve identified how bad the problem is, you can work on solving it. You can explore a number of different solutions for addressing data quality issues, related to people, process, or technology.
In the case of the poor email list data quality example, you have options, including an immediate one to fix your existing problem (bulk email list cleaning) and one that is more long term, avoiding bad emails before they even enter your database (real-time validation at point of capture.).
4. Assess your results
After your solution has been implemented for a defined period of time (e.g. one list cleanse, one month of real-time validation), run another test against your initial metrics. Have your results changed? Was there an improvement in your data quality criteria? Based on what happened, you can adjust or change your solution accordingly.
Data quality can be mean something different from one organization to the next. But as long as you are defining criteria that make sense for your business and testing against those, you can be sure you’ll be able to find ways to drive improvement.
Get a preliminary idea of your data quality through our self-assessment.