Perspectives

Technology & change

A useful system leaves room for judgment

A review step gives a person little control if they cannot understand, question or change the recommendation. Decide what meaningful judgment requires before choosing what to automate.

Keeping a person in a process does not automatically give them control. They may be left to approve a recommendation they cannot explain, manage exceptions they cannot resolve and accept responsibility when something goes wrong.

A useful system should help people make a better-informed decision. That means bringing relevant information together, making assumptions understandable and giving the reviewer authority to act on a concern.

Before choosing the technology, agree on the decision and the responsibility people will retain. A manual approval step cannot settle that question by itself.

Make the review meaningful

A purchasing recommendation reflects choices about demand, stock and supply. The person reviewing it needs to understand enough about those choices to recognize when the recommendation may not fit the situation.

I led the development of store-specific replenishment tools that bring sales, inventory and supply information into purchasing recommendations. People review those recommendations and remain responsible for submission.

I value that distinction. It creates a place for judgment before action. The next question is what makes the review useful: the information available, the time to examine it and a clear way to challenge the recommendation.

Listen before the design becomes fixed

People close to the work can identify unreliable information, important exceptions and consequences a proposed system has missed. Involving them early is a way to understand the decision as well as prepare for change.

Listening does not make every preference a requirement. Leaders still need to make tradeoffs, explain the purpose and decide which concerns change the design. People should be able to see how that reasoning connects to the work they will do.

Training should cover more than the controls. A reviewer needs to know when to look more closely, what they can change and how to raise a problem that belongs to the system. Otherwise, responsibility can remain personal while authority becomes unclear.

Choose where human attention adds value

More automation can be appropriate. Requiring people to inspect everything can create a burden that makes careful review less likely. The choice should reflect the consequences of an error, the reliability of the information and how the business will notice changed conditions.

Be explicit about who owns the system’s purpose and limits, which decisions need human judgment and who responds when the people using it find a problem. These responsibilities continue after launch.

Judge the result by whether the team can understand and act on the business’s choices. If the reviewer cannot question a recommendation or change what happens next, the organization has preserved a step in the process rather than meaningful control.