The Operations Bridge

Post 3 of the five-part series From Data to Decisions: The “What, So What, Now What” Series. We’ve established that a trustworthy fact is the foundation. Now it’s time to ask why the fact matters.

Visibility is not the same as insight

A dashboard can make information visible without making it meaningful.

It can show trends, comparisons, rankings, and exceptions. It can place an important number in a large card at the top of the screen. It can color a result red, yellow, or green.

None of those choices explain why the result matters.

That is the work of the “So What” layer.

“What” establishes the fact. “So What” interprets the fact in context.

This is where reporting becomes analysis.

Context changes meaning

Imagine that an operational measure declines from one day to the next. That is the “What.”

The meaning depends on context.

Was the decline caused by a planned activity? Did it occur during a startup period? Was it concentrated in a short interruption or distributed throughout the run? Did the operating conditions change? Has the same issue appeared repeatedly? Did the result affect the customer, schedule, cost, quality, or only an internal target?

The same numerical movement can represent a serious emerging problem, an understood and controlled condition, or normal variation.

Without context, leaders may react to noise or overlook a signal.

Move from totals to relationships

Static reports often emphasize totals. Insight frequently appears in the relationships between measures.

A few examples:

• Overall performance compared with performance only during active production
• Total downtime compared with the frequency and duration of individual events
• Changeover duration compared with the time required to reach stable performance afterward
• A recurring equipment category compared across operating conditions
• A quality trend compared across positions, time windows, or production runs

These relationships help distinguish symptoms from contributing conditions.

This is one reason interactive analytics became so useful in my own learning. Instead of accepting a single summary, I could filter, compare, and test possible explanations. Power BI allowed the report to become a structured conversation with the process.

But the software did not create the questions. Operational curiosity did.

Ask questions that challenge the first explanation

The first explanation is often attractive because it is familiar.

“The equipment had a bad day.” “The changeover took too long.” “The team did not hit the rate.” “The schedule was difficult.”

Each statement may contain some truth, but none is sufficiently precise to guide improvement.

The “So What” process asks us to go further:

• What evidence supports that explanation?
• What would we expect to see if it were true?
• Do we see the same pattern elsewhere?
• What changed before the result changed?
• Is the issue driven by severity, frequency, duration, or recovery time?
• Are we measuring a cause, a symptom, or an outcome?

These questions do not require advanced artificial intelligence. They require disciplined reasoning. AI and analytics can help explore the evidence, but leaders remain responsible for deciding whether the explanation fits reality.

AI can accelerate interpretation—but it needs boundaries

Working with ChatGPT taught me that AI is very good at proposing possibilities. Given a description of a pattern, it can suggest hypotheses, additional comparisons, statistical methods, or ways to visualize the issue.

That is useful, but a plausible explanation is not a verified explanation.

AI does not stand at the production line. It does not hear the equipment, observe the material, know that a procedure changed, or understand an informal workaround unless someone provides that context. It can reason from the information it receives, but missing context can produce confident conclusions that do not fit the operation.

The appropriate role of AI at the “So What” stage is to expand and sharpen human analysis—not to replace verification.

A productive workflow looks like this:

1. Use the dashboard to identify a meaningful pattern.
2. Ask AI to help generate hypotheses or analytical approaches.
3. Compare those possibilities with process knowledge and source data.
4. Speak with the people closest to the work.
5. Test the explanation against additional evidence.

The result is stronger than either human intuition or AI-generated analysis alone.

Insight should narrow attention

The purpose of “So What” is not to create a longer list of interesting observations. It is to focus attention.

A useful insight tells the team something such as:

• The overall result is being driven by one recurring pattern rather than broad underperformance.
• The visible downtime is less important than the slow recovery that follows it.
• The variation is concentrated in a specific condition and is not present across the entire process.
• A problem believed to be random is recurring under similar circumstances.
• A widely discussed issue is not materially affecting the outcome, while another issue receives little attention.

That changes the conversation. The team is no longer looking at everything. It is looking at what matters.

A question for leaders

When your team reviews a red metric, what happens next?

Do you immediately assign an action, or do you first establish why the result matters and what evidence explains it?

The distinction is important. Action without understanding creates activity. Understanding should improve the quality of the action.

In the next post, we will complete the framework with the question that determines whether analytics produces value: Now what?