EHS Data Analyst in 45 Days: What to Fix Before the First Dashboard Review

A new EHS data analyst often inherits a dashboard that already looks finished. The charts have colors, monthly trend lines, a few targets and a slide that leaders recognize. That polish can be dangerous because the dashboard may be answering the wrong question with numbers that nobody has tested since the template was copied.

The role is not to make safety data prettier. The first job is to protect leadership from false certainty. When a dashboard turns weak definitions, missing context and delayed field evidence into a clean executive view, it can make serious risk look managed precisely when the organization needs better doubt.

This 45-day plan is for the analyst who has to prepare a first dashboard review with EHS and operations leaders. The thesis is practical: data trust is built before the meeting, through definitions, source checks, ownership, weak-signal logic and review questions that force leaders to decide what must change in the work.

Key takeaways

• A safety dashboard should show what leaders need to decide, not everything EHS can count.
• Metric definitions, source ownership and update rhythm should be checked before visual design changes.
• Lagging indicators, SIF precursors, control checks and field evidence should stay distinct because they answer different risk questions.
• The analyst should identify data that is clean because reporting is weak, not because the operation is controlled.
• Dashboard review quality depends on the questions leaders ask after seeing the numbers.

What an EHS data analyst needs to understand before starting

The analyst enters a political system, not only a data system. Every metric has a history. Some numbers became important because a regulator requires them, some because a former leader liked them, some because a bonus formula made them visible, and some because the organization never removed an old chart after the risk changed.

Across 25+ years in executive EHS roles and more than 250 cultural transformation projects, Andreza Araujo has seen that leaders often overtrust indicators when the dashboard is stable and familiar. Stability is useful only when the underlying definitions are stable as well. A chart that compares sites with different reporting thresholds is not a benchmark. It is a source of distortion.

ISO 45001:2018 requires organizations to evaluate occupational health and safety performance, monitor relevant indicators and retain evidence. That requirement does not mean the dashboard should become a warehouse for every number. It means EHS has to know which evidence proves that controls, participation, competence and operational discipline are working.

The analyst should start by asking what decision a competent leader would make differently if each metric moved. If the answer is vague, the metric may be decorative, politically convenient or too poorly defined to belong in the first review.

First week: trace every metric back to its source

During the first week, build an inventory of every dashboard metric. Include the name, definition, formula, owner, source system, update date, reporting threshold, sites included, exclusions, manual adjustments and the meeting where the metric is used. This is not clerical work. It is the foundation for leadership trust.

The common mistake is to start by changing the layout. Better colors can make a weak dashboard more persuasive, which is the opposite of what safety needs. Before touching the design, the analyst should know whether a recordable injury, restricted work case, near miss, critical-control check or SIF precursor means the same thing across all sites.

The recent guide on building a safety metric dictionary is useful here because it separates metric naming from metric governance. A dictionary is not a glossary for auditors. It is the agreement that prevents leaders from comparing different realities on the same slide.

By Friday of week 1, the analyst should be able to mark each metric as reliable, questionable or unusable for leadership review. That label should be based on evidence, not preference.

Days 8 to 21: clean the definitions that change decisions

The second phase is definition repair. Start with the metrics that shape decisions, such as TRIR, severity rate, lost-time cases, restricted work, near misses, SIF precursors, control verification, action closure and overdue high-risk findings. Do not try to clean the entire database at once. Clean the numbers that leaders will use to allocate attention, money and authority.

TRIR and other recordable injury rates can help leaders see part of the injury pattern, but they should not become the whole safety story. A site can have a low recordable rate and still have weak isolation, poor contractor control, failing interlocks or repeated high-potential events. That is why the article on zero-accident target distortions matters for dashboard work.

James Reason's work on latent failures is useful for analysts because it reminds the organization that harm is often prepared before the final event. A dashboard that waits for injuries can miss the conditions that made the injury possible. The analyst should therefore protect separate views for control quality, weak signals and serious-risk exposure.

Definition repair should end with one decision rule per metric. If a SIF precursor rises, who reviews it and within what time? If control verification fails, who owns correction? If near-miss volume drops sharply, who tests whether reporting fell silent? Without decision rules, clean definitions still do not change the work.

Days 22 to 35: compare numbers with field evidence

During days 22 to 35, the analyst should test whether the dashboard matches what supervisors, workers and field verification are showing. This phase protects the organization from a clean data set that hides a noisy operation.

Use triangulation. Compare incident data with control checks, audit findings, maintenance backlog, stop-work events, corrective-action aging, worker concerns, supervisor observations and repeated deviations. The article on indicator triangulation explains why cross-checks reveal risk that a single metric cannot carry.

One practical test is to look for impossible cleanliness. If a high-risk operation reports no near misses, no stop-work events, no critical-control failures and no worker concerns for several months, the analyst should not celebrate first. The better question is whether reporting has become too narrow, too slow or too uncomfortable.

Andreza Araujo's Muito Alem do Zero, translated for English readers as Far Beyond Zero, warns against treating the absence of accidents as proof of maturity. For an analyst, that warning becomes a dashboard rule: silence is a data point that needs interpretation before leaders call it performance.

Days 36 to 45: prepare the first leadership review

The final 10 days should prepare the review meeting. The analyst should not enter the room with only charts. The stronger package is a dashboard, a short data-quality note, a list of metrics that are not yet reliable, 3 decisions leaders need to make and 3 questions that test whether the numbers reflect the field.

Leadership review should separate 4 views. Lagging indicators show what has already happened. Leading indicators show whether planned actions are occurring. Control verification shows whether barriers are present and working. Field evidence shows whether people closest to the work see the same risk picture. When those views are collapsed into one score, leaders lose the ability to decide precisely.

The analyst should prepare uncomfortable explanations in plain language. If one site looks safer because it reports less, say that. If a metric improved because the denominator changed, say that. If an overdue-action chart looks better because low-risk items were closed while serious findings aged, say that as well.

This is where the analyst's credibility is built. The job is not to embarrass operations or defend EHS. The job is to make the decision environment honest enough that leaders can see where risk is escaping the current rhythm.

Month 2 and month 3: turn the dashboard into a governance rhythm

After the first review, the analyst should move from cleanup to governance. Each metric needs an owner, a source, a review frequency, a decision rule and a retirement rule. A metric that never changes a decision should either be moved to a background report or removed from the executive view.

Use month 2 to test whether leaders act on the dashboard. Did the review trigger a field visit, a resource decision, a control repair, a contractor intervention or a change in supervision rhythm? If the meeting produced only explanations and no decisions, the dashboard is still a reporting product rather than a management tool.

Use month 3 to audit leading-indicator quality. Many organizations count safety conversations, inspections, training hours and observations without testing whether those activities find serious exposure. The article on auditing leading indicator quality gives a deeper method for that next phase.

The analyst should also set a retirement rule. Old metrics accumulate because removing them feels risky. In practice, an overcrowded dashboard teaches leaders to skim. A smaller set of better-governed indicators is usually more useful than a crowded slide that nobody challenges.

Common mistakes in the first dashboard review

The first mistake is treating the dashboard as neutral. It is not neutral. It directs attention, rewards some behaviors, hides others and shapes what leaders believe is under control. If the dashboard overweights injury counts, leaders may miss serious-risk precursors that have not yet produced harm.

The second mistake is letting site rankings dominate the conversation. Rankings can be useful when definitions are consistent, but they become harmful when reporting maturity differs. A site that reports more weak signals may look worse than a site that stays quiet, although the first site may be giving leaders a better chance to intervene.

The third mistake is confusing action closure with risk reduction. A closed action can mean a real control was restored. It can also mean the action was administratively completed without testing whether exposure changed. Metric hygiene matters here, because a clean closure rate can hide weak corrective-action quality.

The fourth mistake is avoiding uncertainty. If the analyst does not know whether a number is trustworthy, the review should say so. Leaders can work with honest uncertainty. They cannot manage risk well when uncertainty is hidden behind a confident chart.

Resources to deepen the role

The analyst should study 3 bodies of knowledge. The first is safety metric structure, including lagging indicators, leading indicators, severity, SIF precursors, control verification and field evidence. The second is organizational behavior, because reporting, underreporting and target pressure all shape the data. The third is leadership decision quality, because the value of a dashboard appears only when leaders change what they inspect, fund or stop.

For Andreza Araujo's editorial line, Safety Culture: From Theory to Practice helps connect repeated decisions with cultural maturity. The Illusion of Compliance helps analysts distrust numbers that look complete but do not prove control. Far Beyond Zero helps leaders avoid the trap of treating no accidents as the final evidence of safety.

Teams that want to connect dashboard quality with broader safety culture can use Andreza Araujo's safety culture and leadership work. The useful goal is not a more impressive dashboard. It is a leadership rhythm in which numbers make weak control visible early enough for action.

FAQ

What should an EHS data analyst fix first in a safety dashboard?

The first fix is metric definition quality. The analyst should confirm what each indicator means, who owns the source, how often it is updated and what decision leaders should make from it.

Why do safety dashboards lose leadership trust?

Safety dashboards lose trust when numbers change without explanation, definitions vary between sites, weak signals are missing and leaders cannot connect the chart to field control.

Should TRIR be the main safety dashboard metric?

TRIR can be one lagging indicator, but it should not be the main decision lens for serious risk. Leaders also need SIF precursors, control verification, severity signals and field evidence.

How long does it take to improve safety dashboard quality?

A focused analyst can improve the decision quality of a dashboard in 45 days by cleaning definitions, checking sources, separating lagging from leading indicators and testing the review questions.

Who owns safety dashboard governance?

EHS should own the technical integrity of safety metrics, but operations, HR, maintenance and senior leaders must own the actions that the dashboard triggers.