Five steps to start

  1. Pick one data set: the one causing the most problems (customers, products or invoices). Only one.
  2. Look at it before touching it: run a simple profiling: how many blanks, duplicates and odd formats.
  3. Define a few metrics: the ones matching your real problems, as in which metrics to use.
  4. Fix the worst first: start with errors that cost money or time, following the cleansing process.
  5. Assign an owner: a business person who reviews the figures every month.

You do not need an expensive tool

With a spreadsheet and a few formulas you can measure completeness, duplicates and validity in a table. Specialised tools pay off when volume or the number of sources grows, not before.

Common mistakes

  • Trying to measure everything at once.
  • Cleaning data without fixing the source of the error, which comes back.
  • Leaving data quality to one technical person.

To see where you stand, try the free calculator or the data governance diagnostic.

Quick answers

Does a small business need data quality tools?

Not at first. A spreadsheet is enough to measure completeness, duplicates and validity on a first data set. Tools pay off as volume grows.

Which data set should you start with?

The one causing the most problems for the business, usually customers, products or invoices. Just one at first.

How often should you measure data quality?

Once a month is usually enough at the start, always with the same rule so you can compare progress.

Want to start by looking at your own data? Data Profiling and Quality Issue Management: How to find where your data quality actually breaks before it reaches production, and what to do once it has — the operational complement to your Data Quality Rules. 29 €, one-time payment, instant download.
View product
Full data quality guide, with a free calculator.