I have seen a dozen decks that found margin. None of It reached the actuals.

This is the objection your case will meet, and you should raise it yourself before anyone else does. 

It is also correct. Most executives with fifteen years in manufacturing have sat through several presentations that identified millions in opportunity, approved at least one of them, and never saw the money appear in a monthly result. Their skepticism is not obstruction. It is pattern recognition, and it is accurate. 

Agreeing with it is what separates your case from the ones that came before. 

Why those cases failed 

Go back to the four layers between a loss and a funded decision. 

Losses. The physical events that consume time on the constraint. Named specifically, from your own asset. 

Measurement. A counting method someone else could repeat and land in the same place. 

Translation. Hours at the constraint converted to contribution margin, with the margin figure confirmed by finance. 

Decision. A named owner, a funded action, a date, and the place the result is expected to appear. 

Failed cases perform the third layer alone. 

They arrive with a benchmark, a percentage, and a multiplication. The arithmetic is usually correct. What is missing is any evidence that the losses were observed rather than assumed, any method a challenger could repeat, and any commitment about who will do what by when. 

So the number is interesting rather than actionable. It gets a good reception, occasionally it gets funded, and then it disperses, because nothing in the case ever specified where it was supposed to land. 

The tell 

You can spot a third layer only case in about thirty seconds. 

It cites an industry figure rather than a measurement from the plant in question. It presents a range instead of a number. It proposes a program rather than an action. It has no owner named anywhere in it. And it never says which line of which report will move, in which month. 

Every one of those is a symptom of the same thing: analysis performed at a distance from the asset. 

What holding all four layers actually buys you 

It changes what you are asking for. 

A third layer case asks the committee to believe a number. That is a hard thing to ask, and belief is exactly what a skeptical executive has learned not to extend. 

A four layer case asks the committee to make a decision. The number came from two weeks of counting on a named asset. The method is written down and repeatable. The margin figure came from finance. The action has a cost, an owner, and a date, and there is a specific place the result will show up where anyone can check. 

None of that requires belief. It requires a yes or a no, and it is perfectly acceptable to get a no, because a no arrives with a reason you can work with. 

What you already have 

If you have followed the last five weeks, you have all four layers on one asset. 

You have named categories from your own counting. You have a method you could hand to a colleague. You have hours converted to contribution margin at a rate finance confirmed in writing. And you have three priced actions ranked by cost per hour recovered. 

What remains is assembling it into something a busy executive will read, which is a smaller task than the five weeks that preceded it.  

Your case gets approved. This is where most of them quietly end. 

The month after approval, the plant returns to its normal reporting. The recovered hours flow into the general result and mix with demand changes, product mix, and everything else moving at the same time. Nobody can isolate them. Six months later, when someone asks whether the program delivered, the honest answer is that nobody knows. 

At that point the next request starts from zero credibility, and so does the one after it. 

Declare the landing zone in advance 

The fix is not complicated. It just has to happen before approval rather than after. 

Specify, in the original case, exactly where the result will appear. Which line, on which report, in which month, reviewed by whom. 

Recovered hours on the case packer will appear as run hours at rate on the weekly constraint report, reviewed in the Thursday operations meeting, with the first measurable movement expected in month three. 

Doing this in advance is what makes it verifiable. Doing it afterward is indistinguishable from choosing a measure that happens to look good. 

Publish it every month, including the bad ones 

One view. Hours recovered against forecast, margin recovered against forecast, and which loss categories remain open. 

Publish it whether the month was good or not. 

This is the part people find genuinely difficult, and it is the part that matters. Reporting a month that fell short costs you very little, because everyone who has run anything knows that results are uneven and a short month with an explanation reads as management. Reporting nothing at all costs you the next approval, because silence reads as concealment whether or not it was. 

The scorecard is a small piece of work. It is also the entire basis of your next funding conversation, and the one after that, and it compounds in a way nothing else in this series does. 

What six weeks bought you 

Six weeks ago you had a miss you could not explain and three explanations that could not be priced. 

You now have one asset. One hidden margin figure measured against design on scheduled time. One hourly rate confirmed by finance. A ranked loss profile built from your own crew’s counting. Three priced actions ranked by cost per hour recovered and sequenced correctly. And a one page case with an owner, a date, and a landing zone. 

That is more than most plants have ever assembled about their largest single margin opportunity, and none of it required a system purchase. 

The last instruction 

Book the meeting. 

Do not wait until the plan is complete or until you have covered the whole line. A defended one-page case on one asset beats a comprehensive plan that arrives next quarter, because the one-page case can be decided and the comprehensive plan can only be discussed. 

Thirty minutes with the person who can approve it. This month. 

Your next capital project may already exist inside the assets you own. Go present it.  

You are not running an improvement program. You are buying hours, and there is a price list. The trouble is that almost nobody writes the price list down, so the hours get bought at whatever price the most familiar option happens to charge. 

Two ways to buy 700 hours 

Our packaging line gives away 1,750 constraint hours a year. Suppose the target is to recover 700 of them, worth $4.2mm in contribution margin. There are two credible ways to get there. 

Buy new capacity. A capital request in the range of $12mm. Twelve to eighteen months before first output. Adds fixed cost, headcount, and floor space. Works out to roughly $17,000 per hour of capacity acquired. And when it is finished, the original asset performs exactly as it does today. 

Collect the hours you already own. A program cost in the range of $380,000. First hours recovered inside ninety days. No added fixed cost, no new floor space. Roughly $540 per hour of capacity recovered. And the asset you will still be running either way gets better. 

Both proposals deliver 700 hours. Only one of them has ever been put in front of your capital committee, and it is not the cheaper one. 

Why the expensive option wins by default 

This is not a story about bad executives. It is a story about proposal quality. 

The capital request arrives fully formed. There is a vendor quote, a payback calculation, a specification, an installation schedule, and a name on it. The evaluation process for that document is well established, everyone in the room has assessed one before, and it is straightforward to say yes to. 

The alternative proposal usually does not exist. Where it does exist, it arrives as a program with a duration and no end state, a benefit expressed as a percentage improvement, and no cost per hour figure that would let anyone compare it against the capital request sitting next to it. 

Given those two documents, approving the capital request is the reasonable decision. The committee is not choosing equipment over reliability. It is choosing the only proposal it can actually evaluate. 

The point is comparability, not opposition 

Sometimes the new line is right. Demand may genuinely exceed what the existing asset can deliver even at full recovery. The equipment may be at the end of its usable life. There may be a product the current asset physically cannot run. 

The failure is not buying capital. The failure is buying capital without having priced the alternative, because that decision was made without the comparison ever being available. 

A cost per hour figure makes the two proposals comparable for the first time. It does not decide the answer. It makes the question answerable. 

What this changes about your role 

If you work in maintenance, reliability, or operations, you have probably spent years making the case for your work on the basis of risk, condition, or good practice. Those arguments are true and they lose to capital requests, because the capital request is denominated in the unit the committee uses and yours is not. 

Cost per hour recovered is that unit. It lets you put a proposal on the table that competes on the same terms, in the same meeting, using the same arithmetic. 

This week 

Take your top three loss categories from the counting exercise. For each one, name a recovery action and estimate what it would cost. Rough numbers are fine at this stage. 

Wednesday we turn those estimates into a ranked list, and the ranking will not be the one you expect. 

You have three or four priced actions and a ranking by cost per hour. Now the sequence, which matters more than the list does, because the same four actions in the wrong order produce a fraction of the result. 

The sequence 

1. Stabilize. Stop the repeat failures on the constraint first. Nothing else holds while the same failure keeps returning, and every subsequent gain gets consumed by the recurrence. This is where precision skills go, because most repeat failures on rotating and reciprocating equipment trace back to how the last repair was performed. 

2. Protect. Lubrication and contamination control, so that the stability you just bought does not decay within two quarters. Skip this and you will be paying for step one again next year, on the same asset, with the same people wondering why it did not hold. 

3. Plan. Move work into planned windows. Now the gains are scheduled rather than heroic, and the constraint hours you spend are ones you chose rather than ones that chose you. 

4. Hand over. Operator ownership of the small interruptions. This scales without adding headcount, and it is the only step that keeps working after the program ends. 

Why plants start at step three 

Planning and scheduling is the most visible of the four, the easiest to purchase, and the one with the clearest deliverable. There is a system to configure, a role to fill, a schedule compliance metric to report. Progress is legible from week one. 

It is also the step that disappoints most reliably when it runs first, because planning a failure that keeps recurring simply produces a well planned recurring failure. Schedule compliance climbs, the meetings improve, and the constraint hours do not come back. 

The plants that get the result start at step one, where progress is slower and much harder to see. 

The decay problem 

Nearly every recovery effort produces results in the first quarter. A meaningful share of them are back to baseline within a year. 

That pattern is the reason your executives are skeptical of your next proposal, and their skepticism is earned. They have watched improvements arrive and leave before. From where they sit, the question is not whether your analysis is right. It is whether anything will still be true in eighteen months. 

The hours come back and stay back when three things are in place: the standard is written down, the skill is verified rather than assumed, and the result is measured on a cadence somebody owns. Remove any one of the three and the decay starts quietly, usually within two quarters, and usually without anyone noticing until the number has already gone. 

Put the hold in the funding request 

This is the practical instruction. 

Do not fund the fix and then go back later for the mechanism that holds it. Later does not come, because by then the number has decayed and your credibility with it. 

Written standards, verified skills, and a monthly view with a named owner belong inside the original request. It is a small share of program cost, it is easy to justify when you present it as protecting the investment rather than as overhead, and it is nearly impossible to add afterward. 

This week 

Rank your three priced actions by cost per hour. Then re-sequence them against stabilize, protect, plan, hand over, and see whether the ranking survives. 

Where the cheapest action is out of sequence, the sequence wins. Write one sentence explaining why the top one goes first, because that sentence is going into next week’s funding case.  

Here is the method. It is manual on purpose, it costs nothing, and it produces a number that survives an audit. Automate later if you want. Count first. 

One asset, one sheet 

A single tally sheet at the constraint. Three columns: time, duration, category. Nothing else. 

The temptation is to add fields. Shift, operator, product, root cause, work order number. Resist all of it. Every column you add lowers the completion rate, and a partially filled sheet with eight columns is worth less than a completely filled sheet with three. 

Five categories, matched to what we covered Monday: 

1. Unplanned stop or breakdown 

2. Running below rate 

3. Changeover or startup 

4. Minor stop or short interruption 

5. Quality, rework, or hold 

Count the small ones 

This is the instruction that has to be explicit, because it runs against everything the crew has been trained to do. 

Anything that interrupts flow gets a mark. Including the ninety second clear. Including the jam that one person fixed without telling anybody. Including the reach in to reposition a case that the operator has done so many times it no longer registers as an event. 

These are the ones your system has never seen, and they are the reason the counting exercise exists. If the sheet only captures what the system already captures, you have spent two weeks confirming what you knew. 

Log the rate 

Once an hour, record actual output against design. One line, one number. 

Rate loss is invisible without this, because nothing stops and nothing alarms. It is also, in a large share of plants, the second largest category on the sheet. Skipping this line is the single most common reason a counting exercise underestimates the problem. 

Reconcile weekly 

This is the step that separates a defensible number from a tally exercise, and it is the step nearly everyone skips. 

At the end of each week, total your counted hours and compare against the gap you calculated in week two. Those two numbers were derived completely independently, one from a design rate calculation and one from direct observation, so they act as a check on each other. 

If your counted hours land within roughly ten percent of your calculated gap, your number is defensible and you should stop worrying about it. 

If they do not, the difference is telling you something specific. Counted hours well below the calculated gap usually means the counting is missing a category, and it is almost always rate loss or minor stops. Counted hours well above the gap usually means the design rate is too high, or scheduled hours were overstated. 

Either way you have learned something before you present, rather than in front of the person deciding whether to fund you. 

Write your prediction first 

Before the first shift, write down what you expect to find. Which category will be largest, roughly what share. 

Then count anyway. 

The gap between the prediction and the result is where the credibility comes from. Being wrong in a documented way is far more persuasive to a skeptical executive than being right by assertion, because it demonstrates that the process produced the answer rather than confirming a position you already held. 

Two weeks, not one 

Two weeks is the minimum that spans enough changeovers, product runs, shift patterns, and crews to be representative. 

One week produces a number a skeptic can dismiss in a sentence, and they will, because dismissing it is easier than acting on it. Do not hand them the sentence.  

Everyone in your plant can describe last quarter’s big breakdown. The date, the failure, the recovery, who was called in. It generated meetings, a root cause report, and possibly a capital request. 

Almost nobody can describe the two minute stop that happens twelve times a shift, every shift, and has for two years. It has generated nothing at all, because it has never been recorded. 

On most constraint assets, the second one costs more. 

Where 1,750 hours actually go 

Our packaging line gives away $10.5mm a year. At $6,000 per constraint hour, that is 1,750 hours. Here is a representative split: 

Category	Hours	Value
Unplanned stops and breakdowns	620	$3.72mm
Running below design rate	430	$2.58mm
Changeover and startup	350	$2.10mm
Minor stops and short interruptions	260	$1.56mm
Rework, scrap, and quality holds	90	$0.54mm

Look at rows two and four. Rate loss and minor stops together are 690 hours and over $4mm, and neither one typically appears anywhere in a standard downtime report. 

Why the reporting system misses them 

Your downtime system is not broken. It was designed to explain breakdowns, and it does that competently. It was never designed to price capacity, which is a different job. 

Thresholds. Most systems only capture stops above a threshold, often ten or fifteen minutes. Everything below that is invisible by design. Twelve two minute stops per shift is twenty four minutes a shift, roughly 120 hours a year on a two shift operation, and none of it exists in your data. 

Work order dependency. Events with a work order attached get recorded well. Events cleared by the operator in ninety seconds do not generate a work order and therefore do not generate a record. 

Operator time. Coding a stop takes time the operator does not have during a run. What gets coded is what there was time to code, which biases the data toward long stops and quiet periods. 

Stale reason codes. The category list was built years ago for a different product, a different package format, or a different line configuration. Operators pick the closest available option, which is frequently other. 

Rate is never alarmed. This is the big one. A line running at 480 bags per hour against a 600 design rate is running. Product is shipping, nothing is stopped, and no alarm has fired. There is nothing in the system that treats it as a loss, and yet it is 20 percent of the asset’s capacity. 

Why this matters more than the total 

You already have a total. You calculated it two weeks ago and it is probably right. 

What you do not have is an address. A total tells you the size of the problem. A split tells you where to act, and it is the split that determines whether the money you eventually get funded actually recovers anything. 

Act on the wrong category and you will spend a real budget, generate real activity, and move the number very little. That outcome is the reason executives are skeptical of the next improvement proposal, and it is entirely avoidable. 

What you cannot outsource 

You can buy the total. Several tools, including ours, will estimate a hidden margin figure from data you already have. 

The split has to come from your asset. Every plant’s profile is different, and the difference is the entire point. If your minor stops turn out to be trivial and your changeover losses are enormous, that changes the action, the budget, and the owner. 

Two weeks of deliberate counting on one asset produces that. Nothing else does.  

The method from Wednesday is simple enough that it should work every time. It does not, and the reason has nothing to do with the method. 

The moment a tally sheet appears at an asset, the crew running that asset decides what it is. A measurement, or a performance review. That decision gets made in the first shift, mostly from tone and context rather than from anything anyone says, and it determines whether the next two weeks produce data worth having. 

If they decide it is a review, the sheet will come back clean, cooperative, and useless. Nobody will lie. The small stops will simply not get recorded, because recording them looks like documenting your own shift’s problems, and no reasonable person volunteers for that. 

Say the purpose out loud before the first shift 

Not in an email. In person, to every shift, before the first entry. 

The message is short: 

We are pricing what this asset gives away so we can go get budget to fix it. Nobody is being timed. Nothing on this sheet goes into an appraisal. If the number comes out big, that helps us, because a big number is what gets the fix funded. 

That last sentence does more work than the rest combined. It aligns the incentive. It tells the crew that thorough counting serves them, which happens to be true, because the funded outcome is fewer interruptions on their shift. 

Then prove it 

Stated intent buys you the first three days. What buys you the remaining eleven is acting on something the crew reported. 

Within the first week, take one finding from the sheet and fix it visibly. It does not need to be the biggest one. A guard that has to be removed to clear a jam, a fitting that has been leaking for a year, a tool that lives two hundred feet from where it is used. Something small, something the operator raised, fixed fast, and acknowledged as having come from them. 

One visible fix from operator reported data buys more accurate counting than any policy will. It converts the sheet from something being done to the crew into something being done with them, and the completeness of the data changes immediately. 

The single conversation that ends it 

Somewhere around day four, a manager who was not in the briefing will look at the sheet, see the number of entries, and ask why this line is stopping so many times. 

That conversation, once, in earshot, ends honest counting permanently and it cannot be repaired. The crew will conclude they were right to be suspicious, and every subsequent measurement effort at that asset will inherit the damage. 

So brief the managers, not just the crew. Everybody who might walk past that sheet needs to know what it is and what it is not. The count belongs to the asset, not to the shift. 

Give the data back 

At the end of the two weeks, show the crew the result. The ranked categories, the hours, and the dollar figure attached to the top one. 

Most people who work on a constraint asset have never been shown what an hour of it is worth. Showing them changes how the asset gets run in ways no procedure achieves, and it costs you a fifteen minute conversation at shift handover. 

It also means that when you come back in six months to count again, you will get a better sheet than you got the first time. 

This week 

Brief every shift. Brief the managers. Put the sheet out. Fix one thing they tell you about, fast and visibly. 

Then next week we take the ranked profile and start pricing what it costs to get the hours back, where some of the actions you are expecting turn out to be the most expensive hours available.  

The calculation is one multiplication. Getting the inputs right is where the value is, and where most attempts go wrong. 

The four inputs 

Design rate at the constraint. Units per hour, established as we covered last week, using the lower of the nameplate and the best documented sustained rate. 

Selling price per unit. From finance, for the product family that actually runs on this asset. 

Variable cost per unit. Also from finance. Materials, packaging, and anything else that scales with volume. 

Contribution margin per unit. Price less variable cost. 

On our line: $26.00 less $16.00 is $10.00. At 600 bags per hour, one constraint hour is worth $6,000. 

Why contribution margin and nothing else 

This is the part that decides whether your number survives its first meeting. 

Not revenue. Revenue includes the material you would have had to buy to make the extra units. Using it overstates the opportunity by a wide margin and invites an easy dismissal. 

Not gross margin. Gross margin has fixed manufacturing overhead absorbed into it. But your fixed overhead does not change when the constraint runs an additional hour. You are already paying for the building, the salaried staff, the depreciation, and the utilities base whether the hour runs or not. Deducting that overhead again from a recovered hour charges you twice for cost you have already incurred. 

Contribution margin. Price less the cost that actually varies with the unit. When the constraint produces an incremental hour of output, the full contribution margin drops through to EBITDA, because nothing in the fixed base moved. 

That last sentence is the one to have ready. It is the reason a $6,000 hour is worth $6,000 to EBITDA and not some smaller figure after allocations. 

How to run the finance conversation 

Ask for one thing: price less variable cost per unit, for the product family that runs on this asset. 

If the response is that it depends on mix, ask for a weighted average across the mix that actually runs on the line, or take the lowest margin product and use that. A conservative figure that finance provided is worth far more than an accurate figure you estimated yourself. 

Get it in writing. An email is enough. What you are doing is converting your number from an opinion into a shared assumption, and a shared assumption is almost impossible to attack six weeks later when you present. The people who skip this step are the people who spend their funding meeting arguing about the denominator instead of the decision. 

There is a secondary benefit that is worth as much as the number. Most reliability and operations leaders have never had a substantive conversation with finance about margin. This one is short, specific, and easy to say yes to, and it starts the working relationship the entire funding case depends on. 

The framework this sits inside 

Pricing the hour is step two of five. 

1. Name the hour. Identify the asset that sets the pace, because only its hours carry the full margin. 

2. Price the hour. Design rate times contribution margin. 

3. Count the missing hours. How many hours the asset did not run at rate. 

4. Buy the hours back. Rank actions by cost per hour recovered. 

5. Bank the hours. Prove the result reached the actuals, which is what gets the next one funded. 

The framework does not change. Only the asset and the numbers do, which is why it keeps working on assets you have not met yet. 

This week 

Run the multiplication. Get the email from finance. Write one line: one hour on this asset is worth this many dollars. 

Your maintenance reporting treats every hour as an hour. Two work orders closed, two hours spent, two entries in the completion rate. As a measure of activity that is accurate. As a measure of value it is badly wrong, and the error is expensive. 

The distinction 

An hour lost on the constraint is lost for the whole plant. Output does not recover, the shipment moves, and the margin on that hour is gone permanently. There is no catching it up later, because later the constraint is busy being the constraint. 

An hour lost anywhere else usually costs nothing at all. Assets with protective capacity upstream and downstream absorb the interruption. Work in process buffers it. The line delivers the same output it would have delivered anyway. 

Same labor cost, same work order, same entry in the completion rate. Radically different value. 

Most plants have been making this trade blind for years, not because anyone chose to, but because the reporting system has no field for it. 

What the constraint hour is worth 

On the packaging line, the case packer runs 600 bags per hour at design rate, and contribution margin is $10 per bag. 

600 times $10 is $6,000. 

Every hour that asset does not run at rate, for any reason at all, costs $6,000 of contribution margin. A stop costs it. A slow rate costs it. A changeover costs it. A rework pass costs it twice, once for the bad units and once for the hour spent producing them. 

The reason does not change the price. That is what makes the number so useful. 

What changes when the hour has a price 

The backlog stops being sequenced by age. Most backlogs are ordered by when the work order was raised and how loudly someone has asked. Once the constraint hour has a price, the backlog gets ordered by which asset the work protects, and that ordering can be defended to anyone who challenges it. 

The planned outage argument gets shorter. A four hour planned intervention costs $24,000 in margin. If it prevents an unplanned stop that historically runs eight hours, it protects $48,000. That is an arithmetic problem now, not a debate between operations and maintenance. 

The spare part conversation ends quickly. A $9,000 part that removes a six hour wait is protecting $36,000. Inventory arguments that used to run for months resolve in one meeting once both sides are working from the same rate. 

The overtime shift gets settled. An overtime shift that recovers five constraint hours returns $30,000 against a few thousand dollars of premium labor. Or it recovered nothing, and now you can prove that too, which is equally valuable. 

Why this is the most useful number in the plant 

Everything else in this series is built on top of it. The hidden plant figure is annual and abstract, which makes it good for getting attention and poor for making decisions. An hourly rate is operational. It works on a Tuesday afternoon, in a fifteen minute conversation, about one asset. 

It is also the number that outlives the exercise. Six weeks from now the project is over, and this figure is still sitting in your head, repricing every decision that comes past you. 

This week 

Get contribution margin per unit confirmed by finance in writing. Multiply it by the design rate of your constraint. Write it on one line. 

Then take one decision your plant made this month and reprice it. Any decision. The overtime you approved, the outage you deferred, the part you did not stock. Run it again at the real rate and see whether the answer changes.