The Operations Bridge

Post 5 of 5, closing out From Data to Decisions: The “What, So What, Now What” Series. If you’re just joining, the series followed my experience learning Power BI with ChatGPT while building operational reporting for a manufacturing environment.

What I thought I was learning

When I began this project, I thought I was learning Power BI.

That was true. I learned how data models shape results, how DAX measures respond to filter context, why operating definitions must be explicit, and how visual choices affect the questions people ask.

But that was only part of the development.

The larger lesson was learning how to collaborate with AI on a real problem where the answer mattered.

There is a substantial difference between using AI to generate a quick response and using it to help build an operational capability. The second requires more from the user.

It requires context, precision, skepticism, iteration, testing, and ownership.

AI did not know our operation

ChatGPT knew far more than I did about DAX syntax and common Power BI patterns. I knew far more than it did about the process I was trying to represent.

That created a productive tension.

I could describe the business requirement, but my first description was not always precise enough to become a valid calculation. AI could propose a technically sound measure, but its assumptions did not always match the operating definition. A visual could render successfully while still answering the wrong question.

Progress occurred through iteration:

• I explained the intended business meaning.
• AI translated that meaning into a possible technical approach.
• I tested the result against known operating conditions.
• We identified where the logic or context failed.
• I refined the definition and tried again.

This did more than solve individual technical problems. It taught me how to think across the boundary between operations and technology.

The framework emerged from the work

The “What, So What, Now What” approach reflects the layers I encountered while developing the reporting system.

What: establish the fact. What happened? Is the source reliable? Is the measure defined correctly? Does the dashboard represent the operation as it actually occurred?

AI can help transform data, write measures, identify anomalies, and present information. But a polished output cannot compensate for a weak definition or unreliable source.

So What: establish the meaning. Why does the result matter? Is it a signal or normal variation? What conditions are associated with it? Which explanation is supported by evidence?

AI can accelerate exploration and suggest hypotheses. It can identify relationships that deserve attention. But operational knowledge and verification determine whether those relationships are meaningful.

Now What: establish the decision. What will we do because we understand the issue better? Who owns the response? What evidence will determine whether the action worked?

AI can help evaluate options and structure a response. It should not replace human accountability, especially where decisions affect people, quality, safety, customers, or business risk.

Together, the three questions provide a way to evaluate whether an analytics or AI initiative is producing genuine value.

The risk of stopping at “What”

Many technology initiatives improve access to information but do not improve decisions.

The organization receives faster reports, better summaries, and more attractive dashboards. Employees save time gathering information. Those benefits are real, but they do not automatically change performance.

If the process ends at “What,” the organization has improved reporting.

If it reaches “So What,” it has improved understanding.

If it consistently reaches “Now What,” it has created a decision-support capability.

This is the standard I now use when thinking about AI adoption. The question is not simply whether people are using AI or whether a tool saves time. The stronger question is whether the combination of people, process, data, and AI helps the organization make a better decision or take a more effective action.

AI as a capability multiplier

I could have treated my lack of Power BI and DAX experience as a reason not to begin. Instead, AI lowered the barrier to learning while I remained responsible for the work.

That is one of AI’s greatest values for leaders. It can shorten the distance between curiosity and capability.

But the learning disappears if we outsource all of the thinking.

Using AI as a teacher means asking for explanations, not only answers. It means requesting alternatives and understanding their tradeoffs. It means testing the output, tracing errors, and being willing to say, “That calculation works, but it does not represent the business correctly.”

The goal is not dependence on AI. The goal is increased human capability through collaboration with AI.

A practical test for any AI initiative

When evaluating an AI or analytics project, I now believe leaders should ask:

1. What will become more visible or reliable?
2. So what will we understand that we do not understand today?
3. Now what decision, action, or behavior will improve because of it?

If the team cannot answer the third question, the project may still be useful—but its business value remains uncertain.

The real outcome

I began with a daily production report and a desire to get more insight from it.

I learned to develop in Power BI. I learned to write and troubleshoot DAX. I learned how important data definitions, context, validation, and visual design are to operational reporting.

Most importantly, I learned that working effectively with AI is not a passive conversation. It is an active discipline.

The human brings purpose, context, judgment, and accountability. AI brings speed, technical reach, pattern recognition, and a capacity to help us learn. The value comes from how well those strengths are combined.

Power BI was the project.

Learning to collaborate with AI was the transformation.

And “What, So What, Now What” became the method for turning both into action.