OEE Loss Prioritization Canvas — Frontline Worksheet

OEE Loss Prioritization Canvas — Frontline Worksheet

Use this canvas at the line, in a huddle, or during a rapid kaizen to capture the highest-value OEE losses and choose which to test first. The canvas helps teams estimate impact, check measurement confidence, surface root-cause hypotheses, and commit to a small experiment with an owner and success criteria.

How to use

1. Collect candidate losses from operators, shift logs, downtime trackers, and quality reports.
2. For each loss, estimate Frequency and Duration using the best available data (operator logs, machine counters, or reasonable observation). Record measurement confidence.
3. Estimate Business Impact in minutes lost per day or units lost per shift (use a conservative estimate if data is uncertain).
4. Score Quick-Win potential (how easy/fast to test and likely to reduce loss) and Evidence Confidence.
5. Prioritize by a simple Priority Score (example formula below), pick top 1–3 to run rapid experiments, assign an owner and deadline, and define success criteria.

Canvas fields (table)

Fill one row per loss. Keep entries short — you can expand in a linked experiment plan.

Scoring & prioritization

Keep scoring simple so teams can decide and act. Example approach:

• Impact score (1–10): translate the business impact into a 1–10 scale. For example, rank losses by minutes/day or units/day relative to other captured losses.
• Quick-win score (1–5): how easy and fast is a low-risk test? 5 = very easy, can test in one shift with local resources. 1 = requires capital or long lead time.
• Confidence (1–5): how reliable is the measurement or observation? Low confidence means you may need a short validation step before a full experiment.
• Priority score (example) = Impact score × Quick-win score × (Confidence / 5). Use this to rank items — higher is better.

When to validate measurement first

If Confidence ≤ 2, plan a quick measurement check (one shift of focused observation or a short data pull) before committing resources. Low-confidence items often look big but may not be the right place to spend scarce improvement time.

Experiment plan (simple template)

1. Objective: What are we trying to change and by how much?
2. Hypothesis: If we do X, then Y will improve by Z within N shifts.
3. Test steps: List clear, short steps the operator/owner will run.
4. Data collection: What to measure, who records it, and when (before/during/after).
5. Success criteria: Quantitative threshold or qualitative outcome that signals success.
6. Owner & timeline: Who runs the test and deadline for results.
7. Next action: If success → standardize; if partial → iterate; if no effect → capture learnings and re-prioritize.

Common prioritization mistakes (avoid these)

• Chasing the loudest complaint rather than the largest measurable impact.
• Prioritizing low-impact items with high visibility that demotivate teams when little progress occurs.
• Running big corrective projects before validating measurement or hypotheses.
• Ignoring operator input about feasibility and sustainment.

Suggested immediate next steps

1. Run a 30-minute huddle to capture 6–10 candidate losses using the canvas.
2. Estimate impact and confidence for each; compute priority scores and pick the top 1–2 to test this week.
3. Assign owners, write a one-paragraph experiment plan, and schedule a short follow-up to review results and decide whether to standardize.

Where this canvas fits in a toolkit

This worksheet is a frontline tool for quickly translating OEE insight into actionable experiments. For a broader toolkit, pair it with:

• Standard experiment report template
• Short measurement validation checklist
• Visual dashboard for before/after results
• Operator-facing one-point lessons and standard work

Example image search phrase

oee losses whiteboard

Notes for facilitators

Keep the session tight and operator-led. The goal is measurable learning, not perfect root-cause analysis. Rapid, small experiments that prove (or disprove) hypotheses build momentum.