Defining Data Profiling: Processes and Usage

Examine collected information and spot errors, inconsistencies, and opportunities with data profiling.

Don’t rely on your eyes alone. Partner with data-profiling companies that can inspect your information at scale and give you insights you can use.

Let’s dig deeper into the meaning of data profiling so you can determine if this is an approach you should add to your toolkit.

What is data profiling?

Many companies gather up information. In fact, more than 90 percent of companies say they're spending more on big data solutions every year. But only 72 percent of them have forged a data-driven culture. Blame errors and missed opportunities. Solve those issues with data profiling.

Data profiling involves combing through your information with digital tools to:

  • Verify. Ensure that the data within your tables matches descriptions.
  • Reveal. Discover the relationships between different sources, datasets, and tables.
  • Correct. Spot input inconsistencies (such as numbers sometimes spelled out) that keep you from making clear connections.
  • Parse. Pull from cleaned data and spice up your reports and presentations.

Data profiling begins with discovery. Three types exist.

  • Content: Spot values that are null, incorrect, or somehow unusual. Tap into each data record individually.
  • Relationships: Find out how information connects and intersects. Use your findings to allow for efficient data reuse.
  • Structure: Ensure that your data is formatted correctly and entered consistently.

Data profiling is a bit like housecleaning. Each file is a potential source of error. Your work helps to keep things tidy.

How does data profiling work?

Almost a quarter of all companies can't make big data accessible for end users. If you're gathering information from hundreds (or even thousands) of sources and you're never checking, cleaning, or massaging it, you could be part of this group.

Use one (or several) proven data-profiling techniques, such as:

  • Column profiling. Scan your tables to spot patterns and inconsistencies. Compare multiple columns to inconsistencies and dependencies.
  • Data analysis. Spot relationships between fields, and eliminate or hone connections if the inputs overlap or don't align.
  • Data rule validation. Create firm outlines that dictate how data is collected and recorded.
  • Pattern matching. Find valid formats for your tables and datasets.
  • Table profiling. Identify missing or orphaned records. Examine how columns intersect and duplicate data.

Cleaning up data is critical. Syncari (a company that offers a data profiling tool) says bad data costs companies 15 percent of revenue.

But if you're intimidated by the idea of checking your data by hand, you're not alone. Visual examination of critical data is both time-consuming and inefficient. Try a software provider instead.

Do your data warehouses contain personal data? Find out more about the rules and regulations surrounding this very special and specific type of information on our blog.


Companies Are Failing in Their Efforts to Become Data-Driven. (February 2019). Harvard Business Review.

The Most Common Problems Companies Are Facing With Their Big Data Analytics. Business Intelligence.

The Catastrophic Cost of Bad Data and Where It's All Headed. (November 2019). Syncari.