Bad data does not usually appear once and stay there by accident.
It keeps reappearing because the same business process keeps creating it. In many organisations, data is extracted, corrected, reused, and never properly fixed at source. Over time, workarounds become the real operating model.
That is why bad data is so persistent: the organisation has adapted to it.
A common pattern looks like this.
A team pulls data from a system because it is incomplete, inconsistent, or untrustworthy. They clean it in a spreadsheet, use it for reporting or operations, and move on. The source system is never updated. Next month, the same extract happens again.
At first, this feels efficient.
Then it becomes normal.
Then it becomes infrastructure.
Soon, the organisation depends on unofficial transformations, personal files, local rules, and hidden knowledge that only a few people understand. At that point, the “system of record” is no longer the real system of work.
This is where many organisations lose control.
When data is managed outside official systems:
-There is no single version of the truth.
– Quality rules vary by team.
– Errors are harder to detect.
– Knowledge becomes personal rather than institutional.
– Change becomes risky because no one fully understands the workaround layer.
This is also one reason AI and automation initiatives struggle. If the underlying data is governed through shadow processes, it is very hard to scale trustworthy outputs.
Bad data is often described as a technical nuisance. In reality, it creates hidden operating costs:
– Manual reconciliation.
– Duplicate effort.
– Delayed decisions.
– Rework across teams.
– Lower confidence in reporting and automation.
The business usually pays for these costs every month, even if they never appear on a formal budget line.
How much of your critical data is actually managed outside your core systems?
If key decisions depend on spreadsheets, extracts, and local workarounds, the organisation may be running a parallel data architecture without admitting it.
That is not a minor issue. It is a strategic one.
In the next post, we turn to the standards conversation. The good news is that the building blocks for trusted data already exist.