Process Hazard Analyses (PHAs), audits, incident investigations, and management of change reviews all generate recommendations. The quality of those recommendations and the follow-through to make sure they were completed correctly and timely often determines whether real, sustained improvement occurs — or whether the same issues keep reappearing.
Weak recommendations can cause confusion, waste time and resources, and lead to loss of important program improvement opportunities. Strong recommendations drive meaningful change and help prevent incidents.
Characteristics of Good Process Safety Recommendations
Effective recommendations are:
- Specific – Clearly state what needs to be done, by whom, and by when.
- Actionable – Use strong verbs (e.g., “Develop,” “Implement,” “Revise,” “Install”) rather than vague ones (e.g., “Consider,” “Evaluate,” “Improve”).
- Measurable – Include criteria for success so it’s clear when the recommendation is complete.
- Targeted at the Root Cause – Address the underlying system issue rather than just the immediate symptom.
- Practical and Sustainable – Feasible within the organization’s resources and culture.
Include Hazards and Consequences When Appropriate
Especially in PHAs, good recommendations should clearly reference the specific hazard or scenario they address, along with the potential consequence if the recommendation is not implemented. This helps leadership and reviewers understand the risk context and prioritize actions effectively.
For example, instead of “Add a high-level alarm,” a stronger recommendation might say: “Install an independent high-high level alarm on Tank 101 to prevent overfill leading to a flammable liquid release and potential fire/explosion in the diked area.”
If a qualitative risk matrix or other risk estimation methodology is used, the risk ranking for the evaluated scenario should be included. In some cases, it may also be appropriate to include the estimated risk ranking after the recommendation has been completed to help prioritize the recommendation based on the degree of risk reduction expected.
Assigning Priority to Recommendations
Clear prioritization helps management allocate resources effectively. Common schemes include numerical ranking (Priority 1, 2, 3) or categorical systems such as:
- Regulatory – Required to meet legal or code requirements.
- Policy – Required to meet company standards or internal policy.
- Observation / Good Practice – Important for continuous improvement but not immediately safety-critical.
Priority should be based on the severity of the hazard, the likelihood of occurrence, and the effectiveness of existing safeguards. High-priority items (especially those with potential for catastrophic consequences) should receive immediate attention and tracking. For example, the numerical ranking such as P-1 can be used to assign the estimated urgency for completion of the recommendation, such as immediate to one month (1), six months (2), or one year (3), based on company guidance.
In some cases, interim actions should also be included to reduce the hazard if the time required to complete the recommendation is too lengthy. For example, the design of a new interlock might require 6 to 12 months for completion, but the hazard is still present in the interim, and the recommendation should require interim fixes such as faster design options or administrative controls to manage the hazard while the final solution is in development.
In lengthy reviews, such as a PHA, action items may also be developed for further follow-up and review before the conclusion of the PHA. These might be reminders to get more information for the team prior to the conclusion of team meetings, possible recommendations that the team wants to evaluate further based on additional information, or other notes of interest to the team. These should be clearly labelled and resolved before the end of the review, either leading to completion of the requested action or potentially the development of recommendations to include in the final report.
In some cases, improvement opportunities might be identified that are not relevant to the primary scope of the review, such as a safety item noted during a PHA that is not related to the process hazards under study. These can usually be documented and communicated to appropriate personnel without being included in the report, based on company guidance.
Common Weak Recommendation Patterns (and How to Fix Them)
Here are frequent problems seen in PHA and audit recommendations:
- “Improve training” → Better: “Revise the operator training program for [specific procedure] to include troubleshooting of [specific upset condition] and document completion by [date].”
- “Review the procedure” → Better: “Update the [procedure number] to include [specific missing step] and perform a formal MOC review by [date].”
- “Consider adding an interlock” → Better: “Perform a LOPA on the high-pressure scenario and implement an independent protection layer (IPL) if the risk gap exceeds company criteria.”
- Obscure recommendation: “Address high level in vessel” → Better: “Install independent high-high level switch on TK-201 to prevent overflow and potential release of flammable material to the diked area.”
- Vague recommendation: “Fix the problem with the relief system” → Better: “Perform relief system design verification per API 520/521 for the new operating conditions on the reactor and implement any required changes by [date].”
Practical Tips for Writing Strong Recommendations
- Use the “SMART” framework (Specific, Measurable, Achievable, Relevant, Time-bound) as a checklist.
- Focus on fixing the system, not blaming individuals.
- Include a clear rationale or reference to the hazard/scenario it addresses.
- Assign ownership and a realistic due date.
- Consider hierarchy of controls (elimination, substitution, engineering controls first).
- Link recommendations to existing PSM elements where possible.
- Remember that the person assigned to complete the recommendation is often different from the person who wrote it. Write recommendations clearly enough that the assignee fully understands what needs to be done without additional clarification.
Recommendation Review Process
A formal review step for quality and technical content is highly recommended before recommendations are issued. This review should check for clarity, completeness, technical accuracy, and whether the recommendation actually addresses the identified hazard and consequence. Having a second set of eyes (often a technical reviewer or PHA team lead) significantly reduces the number of vague or ineffective recommendations that reach management.
Recommendations should be reviewed by management to accept or reach agreement that they are relevant and need to be done, in some cases, including revisions to the recommendations for clarity or other reasons. Management should then assign resources and timing to the recommendations for completion and include the recommendations in the facility tracking software. In some cases, spreadsheets or other approaches may be used for tracking, but this is now relatively rare.
Management should also consider if the recommendation should also be applied more broadly to impact facility systems that may be impacted rather than just fixing on thing. For example, a training issue may impact across a site rather than just in the specific area it was identified. Similarly, a fix identified for one piece of process equipment should be reviewed to see if related equipment should be included in the recommendation. A fix for one reactor, for example, might apply to several other reactors as well.
Recommendations should be clearly communicated to affected personnel and, in particular, to personnel assigned for follow-up to complete recommendations. If personnel changes occur, a process for re-assigning recommendations to other personnel for completion should be implemented.
Appropriate metrics for different types of recommendations should be developed and tracked to ensure recommendations are completed on the required timing, with special emphasis on recommendations where the completion is overdue for some reason. Company guidance should be provided on ensuring that metrics are appropriately tracked, trended, and completed as required.
The next scheduled PHA or audit also provides as an opportunity to review the status of previous recommendations to help ensure they were completed correctly and met the specific desired improvement intent.
If an incident occurs that relates back to previous or similar incidents, the current investigation should determine if previous recommendations were completed correctly or perhaps why additional recommendations had not been made that might have been relevant.
Rejecting Recommendations
OSHA has provided guidance on rejecting recommendations: An employer can justifiably decline to adopt a team’s recommendation where the employer can document, in writing and based upon adequate evidence, that one or more of the following conditions is true:
- The analysis upon which the recommendation is based contains material factual errors;
- The recommendation is not necessary to protect the health and safety of the employer’s own employees, or the employees of contractors;
- An alternative measure would provide a sufficient level of protection; or
- The recommendation is infeasible.
Proper Completion of Recommendations
Some companies have approval processes to ensure that recommendations have been completed correctly, meeting the full intent. Proper wording of the recommendation can aid this process.
If a recommendation says to study a problem, it should also state what to do once the study is completed. Closing the recommendation based on the study, but not completion of the results/conclusions of the study, can lead to situations where the true intent of the recommendation has not been met through implementation of an action to actually improve anything.
If a new recommendation is developed based on the study, it can become detached from the original source, such as a PHA, and lose priority for completion while the original closed PHA recommendation suggests that something was done. Closing the recommendation by either issuing an MOC or a new recommendation/action item in the tracking software is considered “closing on a promise” because the new MOC or action item may never be done although the original recommendation is shown as complete.
In some companies, a six month quality check and follow-up on closed recommendations might be used to monitor the effectiveness of the recommendation closures at the facility.
Final Thoughts
Well-written recommendations are one of the highest-leverage activities in process safety. They turn insights from PHAs, audits, and incident investigations into lasting improvements that protect people, the environment, and the facility.
Take a moment to review recent recommendations from your PHA or audit. Are they specific, actionable, and likely to drive real change?
What challenges have you seen with writing or implementing process safety recommendations? Share your experiences in the comments or contact me at jim@psmnews.com.