The hierarchy of controls is one of the most widely taught concepts in EHS. Eliminate the hazard. If you cannot, substitute. If you cannot, engineer it out. Then administrative controls. PPE last.

Everyone agrees with the principle. The challenge is execution.

In real operations, CAPA decisions are constrained by downtime windows, engineering bandwidth, contractor turnover, procurement lead times, budget cycles, and the simple fact that you have to keep operations running. That is why many investigations end in training, reminders, and procedure updates. They are fast. They are easy. They are also often fragile.

This post lays out a five-step practical framework to choose CAPAs that still respects the hierarchy, but also makes tradeoffs explicit by assessing each CAPA option on impact, effort, and confidence, then presenting a balanced bundle of controls for human decision makers.

Step 1: Define the risk in terms of failure modes, not the incident label

Most teams are highly proficient in this step. They don't start a CAPA thinking from “forklift near miss” or “hand injury.” Those labels describe outcomes, not the underlying mechanism.

Instead, they define:

This turns one incident into a prevention opportunity for a whole class of events.

It also aligns with how regulators and guidance documents treat control selection: you are managing risk, not just closing a case. OSHA’s hazard prevention guidance and hierarchy worksheet are practical reminders that control selection is a deliberate process and often uses a combination of controls. (osha.gov) (osha.gov)

Step 2: Build a “control menu” before choosing

Before you debate feasibility, force the team to create options across the hierarchy. This prevents the default-to-training trap.

A practical rule:

This is consistent with how many authoritative sources frame the hierarchy: start at the top, but expect real programs to layer controls, including interim measures while longer-term solutions are developed. (osha.gov) (cdc.gov)

A useful mental model from applied safety writing is: create the menu first, then decide. (thesaferchoice.co.uk)

Step 3: Assess each CAPA option on two axes: Impact and Effort

This is where the hierarchy becomes operational.

Instead of saying “engineering controls are better,” quantify:

Impact score (1–5, or simply High/Medium/Low)

Impact should mean expected reduction in risk, not “how good it sounds.”

Score impact on:

Suggested anchors:

This aligns with the core rationale behind the hierarchy: higher-order controls reduce reliance on perfect human behavior. (cdc.gov)

Effort score (1–5, or simply High/Medium/Low)

Effort is more than cost. It includes time, disruption, and organizational friction.

Score effort on:

Suggested anchors:

This fits well with “reasonably practicable” thinking: real organizations weigh risk reduction against time, trouble, and cost. UK HSE’s framing is explicitly about balancing the level of risk against the sacrifice required to control it. (hse.gov.uk)

Step 4: Add a third dimension: Confidence (important for AI-output)

If you want honest decisions, you need a way to represent uncertainty.

Use a confidence score (1–3) based on:

This stops teams from over-claiming certainty about CAPA effectiveness.

Step 5: Calculate a priority score, then choose a balanced portfolio

A simple formula that is easy to explain:

CAPA Priority = (Impact × Confidence) / Effort

This produces an initial ranking, but do not treat it as a single winner. The best outcome is usually a bundle:

This is consistent with practical regulator guidance that recognizes interim controls while longer-term solutions are planned and implemented. (osha.gov)

Example: Mobile equipment and pedestrian interface

Failure mode: pedestrian exposure in travel lanes and blind spots

Control menu options:

Illustrative scoring:

CAPA optionImpact (1–5)Effort (1–5)Confidence (1–3)Score
Physical separation and controlled crossings5433.75
Proximity detection system4322.67
Traffic rules + speed policy refresh3223.00
Training refresh only2124.00

The math highlights a reality: low-effort actions can look “high priority” because effort is in the denominator. That is why you need governance rules:

This is also why many SIF prevention guides emphasize selecting CAPAs using the hierarchy and prioritizing stronger controls. (ingaa.org)

What a good CAPA decision output typically looks like

A strong CAPA output is not a single action. It is a structured decision package:

  1. Top 3 to 6 options ranked by impact, effort, and confidence
  2. A recommended bundle:
  3. A rationale tied back to failure modes and critical controls

This is what allows leadership to make informed decisions.

Where AI-powered CAPA tools transform the process

The hard part of running this consistently is not the math. It is doing the surrounding work at scale:

Tools in the AI-powered RCA and CAPA category, especially those built around a safety knowledge graph like Haven’s, can help by:

The goal is not “AI decides, but AI provides well-reasoned options and the humans expert decide.

Below is a sample AI recommendation from HavenEDGE CAPA module, which is powered by Haven's Industry Knowledge Graph.

References