Performance visibility: seeing losses while the shift can still act

Why visibility sets the ceiling for the other five pillars

Every other practice in the index runs on information someone captured. A daily meeting can only discuss the stops that were recorded. A Pareto can only rank the reasons operators were able to choose. An escalation rule only fires if somebody sees the stop in time.

When visibility is weak, the rest of the system works hard on a partial picture. Teams argue about whether a loss happened instead of why it happened, and improvement projects target the losses that were easy to record rather than the ones that cost the most.

Timing is the other reason this pillar carries weight. A loss found in next week's report can only be explained. A loss seen within the hour can still be recovered in the same shift, or at least contained before it repeats. Most of the distance between Level 2 and Level 4 here is the distance between explaining a shift and steering one.

The pillar is not about screens. A paper hour-by-hour board filled in honestly by the team scores higher on speed of knowing than a dashboard in an office nobody at the line looks at. We score the practice around the data, whatever tool carries it.

The five levels as you would see them on the floor

The table describes what a plant typically looks like at each level of this pillar. Your pillar index from the assessment places you in one of these rows. The overall level follows the weakest-pillar rule explained on the levels page, so a low score here holds the whole plant back.

The five questions and how plants misjudge themselves

Each question below is answered on a five-point scale. For each one: what we are really asking, where self-assessments usually go wrong, and a check you can run this week before you answer. Answer for your bottleneck line on a normal week, not your best line on your best week.

1. How are stops and losses captured? We are asking whether the record of lost time on your bottleneck is complete enough to act on, and how soon after the event it is written. The usual over-rating is answering 3 or 4 because a terminal or a reason list exists, while stops of a minute or two never reach it.

2. How quickly does the line leader know the line is behind? Knowing means seeing it, not sensing it. Plants over-rate when a live screen exists in the supervisor's office but not at the line, or when an hour-by-hour board is filled in at the end of the shift in one go. The handwriting usually gives that away: same pen, same slant, every hour.

3. Which measure do you use, and how is it defined? OEE, TRS or an efficiency ratio all qualify. The question is whether everyone calculates it the same way. Over-rating happens when lines disagree on planned time: whether breaks, changeovers, preventive maintenance or no-order periods count. Under-rating also happens: plants with a clear written definition of a home-grown ratio sometimes answer 2 because it is not called OEE.

4. Do people trust the numbers? Watch what happens in the morning meeting when a figure looks bad. The trap is answering from the manager's chair. The people who distrust the figures are often not in the room, and they show it by keeping their own counts.

5. Who sees line performance, and in what form? We are asking whether the people who can change the result see it while they still can, in the same terms as management. Plants over-rate when results are posted at the line in a format nobody reads, such as a printed pivot table, or when the line uses different loss categories from the monthly review.

What Level-4 plants do differently

Plants at Level 4 on visibility rarely have more data than others. They have fewer arguments about it, because they set explicit rules for how data is captured, checked and shown. These are the practices that tend to separate them.

• Unexplained time is treated as a defect. No shift closes with stop time that has no reason. The team leader checks it before the handover, and the share of unexplained time per shift has a named owner.
• The capture threshold is a written decision. They decide the shortest stop that must carry a reason, write it in the definition, and group anything shorter as short stops so it stays visible as a block of time rather than vanishing.
• Operators have one data job. The machine signal records when the line stops; the operator only assigns the reason. Asking operators to log start and end times as well is what made earlier systems fail.
• A one-page definition, signed. Planned time, ideal cycle time per product and what counts as a good part, with a worked example. Production, maintenance and finance sign it, and only one named person can change it.
• A 15-minute weekly data check. Counts against shipments or the ERP, recorded time against calendar time, and a written list of corrections with the cause of each. The same errors appearing twice become a problem to solve.
• Hour-by-hour boards the team fills. The team writes the number each hour and a reason next to every missed hour, in their own words. The supervisor initials the board on their walk, which is how you know it was filled live.
• Alert thresholds with a named response. For example, two missed hours in a row, or a stop longer than the escalation limit, brings a named person to the line within a set time. The threshold and the response are written on the same sheet.
• One loss tree from line to plant. The categories on the line board match the ones in the monthly review, so an operator and the plant manager can talk about the same loss without translating.

A 90-day plan to move up one level

Your report gives a specific next move for each question, based on where you answered. The plan below combines them for one typical starting situation: stops logged on paper or in a spreadsheet, figures seen the next day, and a measure that varies between lines. Run it on the bottleneck only. Extending to other lines comes after it holds.

1. Weeks 1–2Name the bottleneck line and the owner of this plan. Write the one-page definition of your measure and recalculate last month with it. Run the unexplained-gap check on five shifts and keep the result as your baseline. Put up a plan-versus-actual board on the bottleneck, even if it is filled at the end of the shift for now.
2. Weeks 3–6Turn the free-text stop reasons from recent weeks into a list of 15 to 25 codes. Move data entry into the shift. Switch the board to hour-by-hour, with a reason next to every missed hour. Start the weekly 15-minute data check and log every correction. If the equipment allows, connect one run/stop signal on the bottleneck machine so stops are detected without anyone writing them down.
3. Weeks 7–12Publish unexplained stop time per shift, with the named owner. Post the shift's losses at the line using exactly the categories management reviews. Write one alert threshold and the response that goes with it. At week 12, repeat the unexplained-gap check and the mid-shift question, then retake the assessment and compare your answers question by question.

Who owns what during the 90 days:

• Production manager: owns the definition, chairs the weekly data check and decides the alert threshold.
• Team leaders: the hour-by-hour board, reasons assigned before handover, and explaining the biggest loss to the next shift.
• Maintenance lead: checks every downtime entry over 30 minutes against the maintenance log for the first month and flags mismatches.
• Controller or finance: co-signs the definition and supplies the good-part count used for reconciliation.
• Engineering or OT: the machine signal and the ideal cycle times per product.
• Continuous improvement lead: builds the reason list with operators and runs the week-12 checks.

Evidence a jury looks for in verification

In verification, the plant uploads evidence and then holds a 45-minute video interview with two jurors. For this pillar, jurors look for artefacts produced by the routine itself, dated and in use, rather than documents prepared for the review.

• Photos of the hour-by-hour board over several consecutive days, showing different handwriting and reasons next to missed hours.
• The written definition of the measure, dated, with the functions that agreed it.
• A raw export or copy of one week's stop log for the bottleneck, showing how much stop time carries a reason.
• Records of the weekly data check: the corrections found and their causes.
• What operators see at the line next to the management report for the same week, so the jury can compare loss categories.
• The alert or escalation threshold sheet as it is posted at the line.
• A trend of unexplained stop time, if you claim Level 5 on the first question.

What does not count: an undated dashboard screenshot, a board photographed freshly written, monthly management slides on their own, or a system specification. A system being installed is not a practice. Be ready in the interview to walk the jurors through one recent shift from the log, and to explain one correction the data check found.

Pitfalls that keep plants stuck on this pillar

• Buying a screen before agreeing a definition. You get faster numbers that are still disputed, and the disputes now happen in front of operators.
• Reason lists that keep growing. Past a few dozen codes, operators pick the first plausible one or 'other', and the Pareto stops meaning anything. Prune the list every quarter.
• Flattering planned time. Excluding changeovers or no-order time until the figure looks good. The number rises and nothing on the floor changes.
• Asking operators to record everything. Start time, end time, reason, part number and comment for every stop. They comply for a month, then the log thins out.
• A percentage without the loss behind it. Showing operators an OEE figure tells them how they did. Showing the three biggest losses tells them what to do.
• Comparing lines on different ideal cycle times. One line uses the nameplate speed, another the best demonstrated rate, and the league table between them is meaningless.
• Handing data quality to IT alone. The errors come from the floor and the fixes are in the routine, so production has to own it.

Where to go next

If you have not yet scored your plant, the FEI assessment takes 12 to 15 minutes and gives you this pillar's index alongside the other five, with a 90-day move for your three biggest gaps. The methodology explains how answers become scores. Visibility gives your teams information; the next pillar, daily management and shift routines, covers how that information turns into action on the same day.

Questions