New Safety Analyst in 30 Days: 6 Decisions Before the First Dashboard Review

A safety analyst does not earn trust by filling a dashboard with more numbers. Trust appears when the analyst can show which measures are defined, which ones are stale, which ones contradict the field, and which ones should never be used as proof that risk is under control.

This role profile is for the person who has just become the safety analyst, the EHS data analyst, or the dashboard owner in a site that already has reports but still lacks confidence in them. The first 30 days should not be spent polishing colors. They should be spent proving that the numbers describe the work, because leaders make decisions from whatever the dashboard makes easy to believe.

Across 25+ years in executive EHS and more than 250 cultural transformation projects, Andreza Araujo has seen the same pattern repeat. The sites that improve fastest are usually not the ones with the most data. They are the ones where someone is willing to challenge a clean number, ask what it hides, and turn the answer into a management decision. In Safety Culture: From Theory to Practice, she argues that culture shows up in repeated decisions, not in declarations. A dashboard is no different.

What the safety analyst needs to understand before starting

The first lesson is that a safety dashboard is not a truth machine. It is a translation layer between field reality and leadership attention, and any translation layer can distort the message if the definitions are loose or the data path is unclear. If the analyst inherits a dashboard that no one can explain, the problem is not cosmetic. It is governance.

The second lesson is that one number rarely tells a safe story. Lagging indicators tell leaders what already happened, while leading indicators are only useful when they sit close enough to the hazard to change action. A green tile can therefore mean either real control or a metric that has drifted away from the work it was meant to represent.

The third lesson is that the analyst is not a graphic designer for executives. The role is closer to a risk translator, because the work is to connect source data, field evidence, threshold logic, and decision rights. If the analyst cannot explain the cause of a number, then the number should not be trusted as a basis for action.

1. Map the data path from source to screen

During the first week, map the full path from source system to dashboard screen. That means identifying where the data starts, who enters it, who edits it, where the formula lives, who owns the threshold, and which report uses the final output. A dashboard often looks simple because the complexity has been hidden upstream.

Use one page per measure. Record the source, the extraction date, the transformation logic, the owner, the intended decision, and the failure mode that would make the number misleading. This step matters because a metric with no named owner can drift quietly for months, especially when the analyst role is shared with another function that is already overloaded.

Andreza Araujo's The Illusion of Compliance is useful here because it warns that paper completeness can hide fragile practice. A dashboard can look complete while the source data is stale, duplicated, or defined differently by each site. The analyst's job is to make that fragility visible before leaders mistake it for control.

2. Write definitions before changing the visuals

In the first 10 days, write definitions before touching the visuals. If the metric is not defined cleanly, changing the chart style only makes the confusion look better. The definition should answer what is included, what is excluded, how the numerator is counted, how the denominator is built, and when the data snapshot is frozen.

This is the moment to build or update a metric dictionary. A metric dictionary keeps everyone honest because it prevents the same number from meaning different things in different meetings. If one site counts a near miss as any unsafe condition and another site counts only high-potential events, the average is not a shared truth. It is a negotiation.

As Andreza Araujo argues in Safety Culture Diagnosis, diagnosis starts with seeing the system clearly enough to ask better questions. The safety analyst should do the same with data. A measure that cannot be defined in one plain paragraph is not ready for executive review.

3. Test data hygiene and contradiction

By day 15, test the hygiene of the data and look for contradiction. Check for duplicate records, missing dates, stale uploads, mismatched categories, and numbers that jump without a plausible operational reason. Clean-looking data can still be weak if it is too smooth, because real work is messy.

Contradiction testing matters even more. Compare the dashboard with at least one field source that can disagree with it, such as supervisor notes, audit evidence, corrective-action aging, or worker voice. If the dashboard says the site is stable while field reports describe rushed work and repeated exceptions, the analyst has found a governance problem, not a presentation problem.

This is where green safety dashboards can mislead leaders if no one asks what would prove the green status wrong. A safety analyst should not defend every green tile. The role is to decide which green tile deserves a second look before it reaches the board.

4. Build a review rhythm that forces decisions

By day 20, the analyst should help build a review rhythm that forces decisions rather than commentary. A dashboard review should answer four questions. What changed, why did it change, what does it mean for risk, and who owns the next move? If those questions are missing, the meeting becomes a tour of charts.

One useful habit is to separate reporting cadence from decision cadence. Weekly reviews should catch drift early, while monthly reviews should reset thresholds, owners, and follow-up actions. If the same issue is discussed three times without a decision, the problem is not the graph. The problem is authority.

During the PepsiCo South America period, where the accident ratio fell 50% in six months, Andreza Araujo learned that metrics matter when they change the rhythm of leadership. A dashboard that only confirms what the team already believes has no operational value. A dashboard that changes a decision is doing real work.

By day 25, translate the dashboard for one specific audience, not for everyone at once. A C-level team needs exposure, trend, and decision signals. A plant manager needs control status and ownership. A supervisor needs a short list that tells them where the pressure is rising this week. A single dashboard can serve all three groups only if the analyst understands that each audience reads the same number differently.

The best way to do this is to create one executive view and one operational view, then keep them linked by the same definitions. The executive view can stay lean, while the operational view carries the detail that helps supervisors and EHS leads act. This avoids the common mistake of forcing the same chart to answer every question, which usually makes it answer none of them well.

The article on plant manager safety scorecard is a useful companion because it shows how leadership uses indicators to see hidden risk. A safety analyst who can serve that need becomes valuable quickly, because the scorecard starts driving action instead of decoration.

5. Protect the metric from misuse

By day 30, the analyst should be able to name the misuse risk for each sensitive metric. Some metrics invite bad behavior the moment they are tied to reward, blame, or reputation. If a number can be gamed by hiding bad news, the analyst has to say so out loud and recommend guardrails.

That means writing the warning next to the measure. Do not use this number as a performance bonus trigger unless reporting quality is independently checked. Do not use this count as proof of safety without field evidence. Do not rank sites only by a green score if the score can be produced by silence. Those warnings are not optional. They are part of the measure.

As Andreza Araujo often frames it, safety is about coming home, not about looking good in a report. A metric that rewards silence teaches the wrong lesson, and the analyst who sees that risk early is doing the organization a service by slowing the meeting down before the wrong decision hardens.

The first mistake is assuming that more data means more insight. It does not. A larger dashboard can hide the signal if the analyst never checks whether the measures are close enough to the work. The second mistake is updating visuals before fixing definitions, which only decorates confusion.

The third mistake is treating every stakeholder as if they need the same story. They do not. A board, a plant manager, and a supervisor need different levels of detail. The fourth mistake is defending the number instead of testing it. If the analyst spends every meeting protecting the dashboard from criticism, the organization will learn that the dashboard matters more than the risk.

6. Resources to deepen the role

Safety Culture: From Theory to Practice helps the analyst connect numbers to repeated decisions. The Illusion of Compliance helps the analyst challenge dashboards that look strong but rest on weak field control. Safety Culture Diagnosis gives the method for comparing declared performance with operated reality. Together they give the role a practical backbone.

If the organization is serious about a dashboard that can stand up to scrutiny, the analyst should work closely with operations, EHS, and leadership rather than sit at the edge of the review process. The job becomes valuable when it helps leaders see the difference between a clean number and a controlled system.

For teams that need help building that discipline, Andreza Araujo's advisory work can support the first dashboard review, the metric dictionary, and the decision rhythm that turns reporting into control. Start with Andreza Araujo.