Data Quality 101 : The Six Dimensions That Are Silently Costing Your Organisation

In a published study, the Gartner Institute suggests that poor data quality costs organisations an average of $12.9 million annually. Also, in 1999, NASA lost a $125 million Mars orbiter because one engineering team used metric units, while another used imperial units.

A simple data discrepancy led to one of the most expensive typing errors in history. These examples clearly show how much unqualified data directly and badly affects the process of decision-making.

In this module, you will learn what exactly is data quality, and understand the hidden costs of poor data quality management.

The Invisible Problem

Most of the time, data quality is invisible until it becomes a crisis. For example, imagine a company that is launching a brand-new marketing campaign based on client databases where the majority of the clients are duplicates.

The marketing campaign has high potential to fail, as it targets only half of the customers, and at the same time, messages are sent multiple times to the same clients!

As a result, the enterprise will lose precious time trying to manually compensate duplicate errors, while taking wrong decisions at the same time, because deciders can’t rely on valuable data to plan the next steps.

In addition, the organisation will lose customers’ trust as workers are unable to match clients’ expectations.

The Six infinity Stones of Bad Data

In the Marvel Universe, villain Thanos spent years hunting down six Infinity Stones — each one powerful on its own, but truly devastating when combined. Poor data quality works in a strikingly similar way.

There are six dimensions of bad data, each capable of causing real damage independently. But when they appear together within an organisation's processes, the consequences can be nothing short of catastrophic.

• Accuracy – A client recorded with two different spellings (MacDonald / McDonald).
• Completeness – Empty fields in the database (missing email, date of birth…)
• Consistency – Different used names for the same product by different systems (iPhone 15 Pro Max/ Apple iPhone 15 PM)
• Timeliness – Non updated postal address
• Validity – Postal Code has 4 numbers instead of 5
• Uniqueness – The same client is recorded three times in the same database

Where does Bad Data Come From?

Poor data quality rarely appears out of nowhere. In most organisations, it builds up gradually, thanks to everyday processes and habits that go unchallenged for years. Understanding where bad data originates is the first step towards addressing it.

Human beings make mistakes; the more data is entered by hand, the more errors accumulate. A misspelt name, a missing digit, an incorrect date. Individually, these seem minor.

Most organisations rely on multiple tools and platforms that were never designed to speak to one another. When data moves between a CRM, an ERP and a marketing platform, something almost always gets lost or distorted in translation. The same customer can end up with three different records across three different systems.

Data has a shelf life. A customer's address, job title, or contact details that were accurate two years ago may be completely wrong today. Also, when there are no guardrails at the point of data entry, anything goes. Free-text fields, inconsistent formats, missing mandatory information:

Lastly, organisational silos is perhaps the most regular source of all bad data. When teams work in isolation, they develop their own definitions, formats, and standards for the same data.

The concept of Data debt

Over the years, organisations put a name to describe poor data quality management and the process of remediation: Data Debt

Like the state debt, data debt is the long-term cost that accumulates when data quality issues are ignored, deferred, or patched over rather than properly addressed. Every duplicate record left uncleaned, every validation rule never implemented, every outdated entry left to linger in your database — these are all deposits into a debt that will eventually need to be repaid.

Acknowledging the existence of data debt is, in itself, a positive sign — it reflects an honest admission that bad data exists within the system and that something must be done about it.

Addressing it requires deploying both financial and human resources, starting with the fundamentals : implementing validation rules and establishing common data standards across the organisation.

In the next module, we will detail how to address data debt across organisational processes.