Most leaders approached at an INSEAD forum refused to bet their own bonus on the completeness and accuracy of company data. Yet the problem often does not originate in the database but in the everyday work of the people who create the information. If they do not see the consequences of a late or incorrect entry, the company simply absorbs the costs of poor-quality data. AI will not solve this problem; instead, it increases the importance of clear business ownership of data.
Data quality can easily become a technical topic. As soon as duplicates, missing fields or conflicting values appear, the problem is handed to the data team or IT. INSEAD points out that this type of response often starts too late.
At the 2026 Alumni Forum in Oslo, leaders were asked a simple question: would they bet their entire bonus on the accuracy and completeness of the data used for the company’s most important decisions? Most declined, a significant share did not know the quality of the data, and only four respondents would have accepted the bet.
Poor-quality data do not have to cause one visible collapse. Organizations often absorb the cost continuously. One expert in the discussion gave the example of a salesperson who spends three hours of a ten-hour workday preparing data that should already have been available.
Such a problem did not primarily arise in an analytical tool. It is a consequence of the process: someone failed to create the information on time, recorded it differently or did not know who would use it later.
INSEAD cites an experience from HelloFresh where even simple information sharing between teams helped. The people creating the data understood the consequences of errors for downstream users, and quality improved.
For management, it is therefore more useful to start with a specific business problem than with a company-wide “data cleansing” program. Choose a decision whose quality is visibly damaged by poor data. Identify which data it needs, who creates them, who uses them and where their meaning or format changes along the way.
Ownership must remain in the business. The data team can set standards, controls and technical mechanisms, but it does not automatically know the business meaning of every field.
AI increases the importance of this discipline. An automated system can process bad data faster and at greater scale, but it cannot by itself repair an unclear business process that creates those data.
The most important question is therefore not whether the company has a modern data platform. It is whether the person entering a piece of information at the start of the process understands who will use it and for what decision at the end.
KEY TERMS
- Data quality: The accuracy, completeness, timeliness and usability of data for a specific business purpose.
- Data owner: The person or function responsible for the business meaning and quality of a specific set of data.
- Data producer: A person or process that creates a data point during ordinary work.
- Hidden data cost: Time and errors caused by correcting, searching for and verifying poor-quality information.