The bill for poor data appears nowhere in the budget
Why the cost of poor data has no address, and how to give it one.

Organisations love doing the maths. Every investment gets a business case, every department a budget and every euro a cost centre. Except for data.
At the Gartner Data & Analytics Summit in London in May, analyst Anurag Raj called data governance the single point of failure for AI ambitions. According to Gartner, missing meaning and context in data lead to inaccurate AI agents and wasted budget. Awareness of the importance of reliable data may be growing, but I can't shake the impression that people are once again looking for the cause in technology or project delivery.
So where does the bill for poor data actually appear in the budget?
The cost has no address
Poor data doesn't send an invoice. The cost lands with customer service as extra calls, with finance as correction work and with sales as missed opportunities. Every department books it under "normal work".
Because it simply comes with the job, nobody adds up the total. And so, for management, it doesn't exist.
Four hours of capital letters, every Monday morning
An example from one of my own assignments. Every Monday morning, an employee changed all the customers' initials back to capitals by hand. Four hours of work, every week. She had learned this during her onboarding, as a fixed part of the job.
It had been going on for years, so it kept going. Nobody asked why the initials kept turning lowercase.
The cause was a small bug in an integration script that ran every weekend. One script, one bug, years of manual work.
The better the catch, the less visible the problem
That employee did her job well. She fixed the error before it really became visible, so there was no complaint and no incident. It probably started that way once: customers complained and management responded with "Just sort it out".
After that, the reports showed correct initials. Management had no reason to ask where that correctness came from.
Craftsmanship (or manual work) hides the error from whoever holds the budget.
Noticeable, never visible
The impact of data quality is always noticeable and never visible. Employees notice that impact in their daily work. Support functions (finance, BI, data teams) clean up the reports. Management doesn't see the impact.
A problem like that gets no place on a dashboard, in a business case or in a budget. And without a place in the budget, it gets no solution either.

Do the maths
Four hours a week is well over two hundred hours a year, more than five working weeks. For one field, with one employee.
Put an hourly rate on that and you have an amount that appeared in no budget at all.
A bug in a script usually takes a day to fix. The manual work went on for years.
You can feel good data too, without seeing it
The reverse also holds. With good data, the impact stays noticeable. The reports add up, Monday morning is free for other work and customers see their own names spelled correctly.
You just don't see it. Nobody writes a memo about an error that doesn't happen.
Invisible here means it works. That is the gain from adequate data: the organisation runs smoothly without anyone having to take an extra step.
For AI, good by accident is an assumption
That gain is on shaky ground if the data is only good by accident. An alert colleague, a clever habit or a script that happens to run properly keep the quality in place, while nobody has recorded that knowledge.
An employee fills the gaps with common sense. An AI agent fills them with assumptions, and those are sometimes wrong. Gartner links that lack of meaning and context to inaccurate agents and wasted budget.
The capital letter rule lived in the onboarding instructions and in the head of one employee. An AI agent doesn't learn capital letters during onboarding.
Give the cost and the quality an address
Ask each department which corrections they routinely make by hand. Have them add up the hours and put the total in the budget, with a name next to it.
Also record what is going well now and, above all, why! As with many things in life, the trick after that is to keep a finger on the pulse. It only takes one small error slipping through unnoticed and poor data creeps back in from another corner. That finger on the pulse belongs at the source, so the error stops there.
Data doesn't have to be perfect. It has to be adequate.
Written by Ruud Kuil
Co-founder of MAD-Quality. Helps leadership teams move from data chaos to data they dare to manage by. Want to know more about data quality? Follow Ruud on LinkedIn.



