If quality is free, why isn't data?
In 1979, Philip B. Crosby introduced a revolutionary concept: “quality is free.” Originally applied to manufacturing, this principle holds profound relevance in today’s data-driven world. The idea is simple yet powerful – investing in quality upfront provides a significant saving because it eliminates the need to fix problems after they occur.
Crosby along with other quality gurus like W Edwards Deming, Joseph Juran, Walter Shewhart and Armand Feigenbaum revolutionised the way we think about quality. They made us realise that building systems, processes and procedures to ensure quality is built in at the outset is far more cost effective than correcting mistakes once made.
How about data quality? Could the same argument be made – in essence that investments in improving data quality are free, because they deliver far greater benefits than they cost?
Of course, the answer is yes – “data quality is free” – but let’s unpack this a little by looking at what we know about the cost of bad quality data.
What do we know about the cost of bad quality data?
There’s a nice rule of thumb, developed by George Labovitz and Yu Sang Chang, called the 1-10-100 rule. Their rule says that if it costs $1 to check the quality of data at source, it costs $10 to clean up the same data and $100 if bad quality data is used. Couple this with the results of a study published in the Harvard Business Review which finds that only 3% of companies’ data meets basic quality standards!
Authors, Tadhg Nagle, Thomas C. Redman and David Sammon, propose an interesting (and simple) exercise to measure data quality. They suggest calculating a Data Quality (DQ) score by taking 100 data records and working through them, marking obvious errors. The DQ score is the number of records that are classified as error-free. In their study a sample of 75 executives who completed this exercise found that “47% of newly created data records had at least one critical error, and only 3% of the DQ scores could be rated “acceptable” using the loosest-possible standard”.
These are scary statistics. If only 3% of data quality scores are acceptable and it costs 100 times more if bad quality data is used, there’s clearly scope to make significant improvements and cost savings by investing in improving data quality at the outset.
What do we know about the investment required to eliminate bad quality data?
Let’s flip the argument and ask – what do we know about the investment required to eliminate bad quality data? Where would we start?
If we apply the logic of the quality gurus, then the trick is to build quality in at the outset. This means putting in place systems, processes and procedures to eliminate bad quality data from the start. A useful way of thinking about this is to consider four issues.
- Establishing Robust Data Governance: Creating clear policies about data ownership, standards, and management.
- Implementing Automated Validation: Using tools that automatically check data integrity at the point of entry.
- Creating a Single Source of Truth: Centralising data management to minimize discrepancies across systems.
- Monitoring Continuously: Using real-time monitoring to catch and correct issues before they propagate.
The bottom line
Investing in data quality isn’t an expense—it’s a strategic investment. By treating data as a critical asset and building quality into every stage of its lifecycle, companies can unlock unprecedented potential and competitive advantage.
In the digital era, quality data isn’t just a technical requirement; it’s the foundation of intelligent business strategy.
If you’ve enjoyed this blog and would like to receive notifications of future blogs please subscribe for updates below.