On-Site Validation Is a Relationship

If someone is coming to shadow your work, here is what that should mean.

Someone from the outside is coming to validate the data. You have heard this before. It usually means extra meetings, screenshots requested at your busiest hour, and a report that describes your job in a way you would not recognize.

This is not how it should be done. There is a better way—one where the validator works collaboratively, respects operational demands, and minimizes disruption to the workflow. This is what I commit to when I conduct on-site validation in a department.

Your performance is not what is being evaluated

The report is being tested. The vendor output is being independently validated. The claim that a number represents operational reality is being examined.

Not your performance.

I am there because a number left the building and told leadership a story. My job is to determine whether the workflow that produced it supports that story. The people who perform the work know what actually happens between the documented process and operational reality.

Shadowing means I work around you

Observation is scheduled around the department’s operational reality, not solely around my availability. If Tuesday afternoon is when the queue is manageable, that is when I sit with you. If the day falls apart, we stop.

I batch my questions instead of interrupting you in the middle of an account. I take my own notes rather than asking you to write things up. I do not ask for a demonstration of something I can observe naturally, and I do not ask you to repeat work you have already completed just so I can see it twice.

Your bandwidth is a real constraint, not an obstacle to my work.

I see this pattern often: staff are already surfacing legitimate errors while working an account or responding to a patient complaint, but they do not have the capacity to determine whether those errors are isolated incidents or evidence of a systemic pattern.

The gap is measurement, not detection.

Using their remaining hours to conduct my investigation would place more pressure on the same operation I am trying to understand.

The workaround is the finding

Every operation has steps that live outside the documented process. A field someone fills in because a downstream report breaks without it. A status that means something different than its label. A spreadsheet that exists because the system will not do what it was supposed to do.

People often apologize for these workarounds.

Please do not.

Undocumented human judgment is often some of the most valuable evidence in the building. It is also the part no vendor report, dashboard, or model can see on its own.

When a reported number and operational reality disagree, the explanation may be sitting in something a person had to decide at 4:50 on a Thursday—or in a system behavior that forced that decision.

Data can be technically correct and still tell the wrong financial or operational story. Finding that gap requires the people who know where the story bends.

What I will ask and what I will not

I will ask what you check before you trust an account. How you interpret the account history. When you escalate and to whom. What the vendor or system does that you have learned to watch for. Which exceptions you handle so routinely that no one has ever written them down.

I will not ask you to evaluate a coworker. I will not time you or create an employee productivity scorecard.

I do not conduct employee performance evaluations, and individual observations are not reported as employee scorecards. Before the work begins, I make that purpose and boundary clear to both leadership and staff.

The workflow facts are checked before conclusions are finalized

This is the part that matters most and the part many reviews skip.

Before I finalize a conclusion, I walk the relevant workflow facts back to the people who perform the work.

If I have misread a step, you correct me. If I have missed a system limitation, you show me. If a pattern has a legitimate operational explanation, that context becomes part of the finding rather than a footnote added after the report is written.

This does not mean employees are asked to approve, reject, or suppress a conclusion. It means they have an opportunity to correct factual misunderstandings, identify missing context, and explain the operational reality behind the data.

That is not a courtesy. It is a validation control.

A finding that has not survived contact with the person who knows the workflow is not a finding yet—it is a hypothesis with a chart attached.

The system, not the person

Independent review works when the people who perform the work are treated as primary sources, not as subjects of the investigation.

That is the premise of the Human Reality Check™: validate data against real human workflow before it influences a dashboard, a board presentation, a financial decision, or a model.

I shadow the work to understand the system surrounding the person—not to point a finger at the person doing the work.

The evidence is better that way. So is the experience of the people who have to keep the operation running while it happens.

JB Human Analytica validates healthcare data against real human workflow before it influences decisions. If you are weighing an independent review and want to know whether the approach fits your operation, request a 20-minute Fit Check at jbhumananalytica.com.