A capex request has a format every board has seen a hundred times: the quote, the payback, the NPV, the vendor’s slide. A capex-avoidance case has no established format. So it tends to arrive as an OEE chart with an arrow on it, and it loses to the quote, not because it is wrong but because it does not look like a decision.
The new-line cost article made the comparison between building and recovering. This one is about the document: the structure that survives a board, in one page, with the proof attached and the fallback stated.
Four reasons, and all four are format problems rather than substance problems.
They are argued in OEE points. The board does not know what a point is worth, and the SKU-level value article explains why nobody else does either until the SKU economics are applied. A case in points asks the board to trust a unit it cannot price.
They present one number. A single value with no range invites the one question that kills it: “what if you’re wrong?” A case with no answer to that question has no answer.
They ask for belief instead of a test. “We can improve by twelve points” is a claim. A board that has heard improvement claims before, and has watched most of them arrive late or not at all, discounts the claim and approves the line, because the line at least exists.
They never say what happens if it fails. The capex case has a built-in fallback: the line gets built and the plant has more capacity. The avoidance case, as usually presented, has none. If the improvement does not arrive, the plant is a year behind. The board can see that even when the presenter cannot.
Each is fixed by structure. Dollars instead of points, a range instead of a number, gates instead of belief, and a fallback stated up front.
Seven blocks, in this order, on one page. Everything else is appendix.
1. The decision. One sentence. “Defer the line-7 purchase by twelve months and fund a $0.4M capacity recovery program across the six existing lines, with a go/no-go at day 90.” The board should know what it is approving before it reads anything else.
2. The measured baseline. The honest OEE on a stated basis, per the OEE audit article. The loss Pareto covering at least 80% of lost time. The split into recoverable through discipline, recoverable through targeted fixes, and structural, from the hidden capacity method. Three lines, with the measurement period and method named. This block is what distinguishes the case from a hope.
3. The two options, side by side. The compressed version of the cost comparison table: cash required, time to first output, time to full output, added capacity, cost per unit of capacity, risk, reversibility. Six rows, two columns.
4. The value, in achievable margin. Not production value. The recovered capacity valued at contribution margin, applied only to the volume the plant can use, with the use stated: this share displaces overtime, this share fills demand, this share is idle. The three-number ladder from the value article, with the third number in bold.
5. The sensitivities. The three variables that move the value and the range they produce. One chart.
6. The proof plan. Three gates, each with a date, a measurable threshold and an owner.
7. The fallback. What triggers the line order, what the delay will have cost, and what the plant will know about the line it buys that it does not know now.
A CFO will test six numbers. The paper should get to each of them first.
Recoverable points. Show the bucket logic and show that only the discipline and targeted-fix buckets are counted. The structural bucket is excluded by construction, and saying so is what makes the recoverable number credible. “Twelve of forty-two points, excluding six structural and twenty-four accepted” is a sentence finance can audit.
The demand use. What share of recovered volume displaces overtime, returns co-packing, fills open demand, or idles. The honest version includes the idle share at zero. A case that assumes every recovered unit is sold will be asked to prove demand it does not have.
The margin. Contribution, not gross. State the variable cost basis. Finance will check it against the costing file, and it should match.
Program cost, including internal time. The CI lead’s half-year, the operators’ hours in loss reviews, the engineering time on the targeted fixes. A program costed at hardware and software alone looks cheap and reads as naive.
The ramp. Gains arrive over months three to nine, not on day one. Show the curve. A case that books the full annual value in year one will be corrected downward by the room; a case that shows the ramp will be trusted on the rest.
The cost of being wrong. If the gates fail and the line is ordered at day 90, the plant has spent one quarter of overtime it would otherwise have spent anyway, plus the program’s sunk cost, in exchange for a line specified against a measured bottleneck. State that cost. It is smaller than the board fears, and stating it is what lets the board approve the test.
Three variables carry almost all the uncertainty, and the paper should show them as a range rather than a point.
Recoverable points: low, base, high. The base is the measured bucket analysis. The low case assumes the targeted fixes deliver half of what the Pareto suggests. The high case assumes some of the accepted losses turn out to be recoverable, which the bulk-to-pack article found on the series line.
Achievable share of recovered volume. The base is the current demand picture. The low case assumes demand softens and more of the recovered capacity idles. The high case assumes the commercial team fills it.
Timing. The base is first gains at month three, full at month nine. The low case slips each by a quarter.
Draw the tornado. The point for the board is that even the low case usually beats the new line’s year-one output, because the line has its own ramp and its own year-one OEE, which the cost comparison article shows starting far below the plant’s 58% and settling there, never at the vendor’s 85%.
And say where the case breaks. If the measured recoverable points are below about six, or if demand is genuinely 30% or more above the lines’ effective capacity, the line wins, and the paper should say so in its own sensitivities section. A case that names the conditions under which it is wrong is the only kind a board should approve.
Ninety days, three gates. Each gate has a number, a date and a name against it. This is what turns “trust us” into “test us”, and it is the reason boards approve avoidance cases that are structured this way and reject the ones that are not.
Day 30: the baseline holds. Machine-verified measurement on the pilot line confirms the OEE within two points of the baseline stated in block two, and the top five loss reasons cover at least 80% of lost time. If not, the Pareto the case was built on was wrong, and the case is withdrawn before any program money beyond the measurement is spent. Owner: the plant manager.
Day 60: the interventions move the target losses. The three highest-value interventions from the Pareto are in place and the specific losses they address have moved by the predicted amount: for the series line, changeover median at or below 30 minutes, case-packer stop rate down 50%, fix-held rate above 80% on the case packer’s repeat failures. Not OEE yet; the losses. Owner: the CI lead.
Day 90: capacity is recovered. At least six of the twelve recoverable points are measured on the pilot line against the day-30 baseline, which is about what the Saturday shift delivered, and the overtime on that line has stopped. The board decides: continue the program across the remaining lines, order narrow equipment at the measured bottleneck for the structural points, or order the line. Owner: the COO.
Each gate is a decision point, not a status update. The board approves the first thirty days and the right to continue; it does not approve a year on faith.
If the gates fail, the line is ordered at day 90. The paper should say so in those words.
Then it should say what the delay cost: one quarter of the overtime the plant is already paying, which on the series plant is around $0.4 million, plus the program’s measurement cost, which is a fraction of that. And what the plant gained for it: a line specified against a measured bottleneck rather than a demand forecast, with the loss profile written into its acceptance criteria, that will start at the effective OEE rather than the demonstrated one because the losses it would have inherited have been found.
The avoidance case is stronger when it admits the line might still be needed. A board that hears “we might be wrong, here is what it costs if we are, and here is why we will have bought a better line” approves the test. A board that hears “we will definitely improve” has heard that before.
The series plant’s paper, compressed to its seven blocks. Figures from the earlier articles in this series; illustrative, and consistent with them.
Decision. Defer the seventh line by twelve months. Fund a $0.4M recovery program across six lines. Go/no-go at day 90.
Baseline. Line OEE 58% on planned production time, measured over four weeks by sensor. Top five losses cover 82% of lost time. Of 42 lost points: 12 recoverable (7 discipline, 5 targeted fixes), 6 structural, 24 accepted.
Options.
| New line | Recovery program | |
|---|---|---|
| Cash | $6 to 8M all-in | $0.16 to 0.45M |
| First added output | Month 9 to 14 | Month 3 |
| Full added output | Month 18 to 26 | Month 9 |
| Added capacity | ~835k units/week at year-one OEE | ~1.04M units/week across six lines |
| Reversible | No | Yes, at any gate |
Value. Per line: production value $3.9M, contribution margin $1.5M, achievable margin $0.8M, of which $0.28M is overtime displaced and $0.51M is demand filled on SKUs A and D. Across six lines, base case achievable margin about $4M a year, ramping from month three: $1.65M of overtime displaced and about $2.4M of demand filled, because the open demand on A and D is a plant-wide figure the lines share, not six times one line’s.
Sensitivities. Low $2.5M (half the targeted fixes, softer demand, one-quarter slip), base $4M, high $5.5M (two accepted points prove recoverable). Case breaks below six recoverable points or above 30% demand growth. Neither is indicated by the data.
Gates. Day 30: OEE within two points of 58%, top five at or above 80%. Day 60: changeover median at or below 30 minutes, case-packer stop rate down 50%, case-packer fix-held above 80%. Day 90: at least six points recovered on the pilot line, Saturday overtime stopped on that line.
Fallback. Order the line at day 90 if the day-90 gate fails. Delay cost about $0.4M in overtime already being paid. Line specified at the measured case-packer and bulk constraints, with the loss profile in the acceptance criteria.
That is one page. The appendix carries the Pareto, the bucket analysis, the SKU economics, the sensitivity model and the full cost comparison, for whoever wants them. Most board members will not read the appendix. They will read the seven blocks, and they will ask about the gates.
The paper above is only as strong as block two and block six, the measured baseline and the proof plan, and those are the two blocks a plant cannot write from existing reports. The pilot and the platform are built to produce them.
The pilot readout is the appendix. Six weeks on one line produces the measured baseline on a stated basis, the loss Pareto with its buckets, the recoverable capacity valued as production value, contribution margin and achievable margin, and the three interventions with owners. That is blocks two and four, and most of the appendix, as a system output rather than a slide deck someone assembles.
The financial impact module supplies the dollars and the ranges. With the selling price and margin entered once per SKU, the recovered capacity is priced by SKU and by use, and the three sensitivity variables can be flexed in the module rather than in a spreadsheet that diverges from the data.
The gates measure themselves. Day 30, 60 and 90 are read from the OEE module on the same sensors that produced the baseline, so there is no second data collection and no argument about whether the gate was met. The changeover median, the stop rate and the fix-held rate are live numbers.
The interventions are a tracked backlog. The AI actionable insights carry the three interventions as items with owners, predicted effect and measured result, so the day-60 and day-90 reports are generated from what actually moved.
The scheduler shows the capacity the plant will run to at each gate. Demonstrated rates and the plan built on them, before and after, so the board sees the recovered capacity as a schedule it can make, not only as a ratio.
The program’s first line item is fixed. The pilot fee is fixed and credited in full against a first-year subscription if the plant rolls out, which means block three’s program cost starts with a known number and the day-30 gate costs the plant the fee and nothing else if the case is withdrawn.
The platform does not write the paper. It makes blocks two, four and six true, which is the part a board cannot otherwise check.
How do you build a business case for OEE improvement? State the decision, the measured baseline with recoverable and structural losses separated, the build-versus-recover options side by side, the value in achievable contribution margin with the demand use stated, the sensitivities as a range, a proof plan with dated gates, and the fallback if the gates fail. One page, with the measurement and the model in an appendix.
What is capex avoidance? Meeting a capacity need by recovering output from existing assets instead of buying new ones, so that capital expenditure is deferred or avoided. The case for it has to be made in cash and risk against the capital alternative, with a test plan, because the alternative has a known cost and the recovery has to earn the same credibility.
How do you present capacity recovery to a board? As a decision with a test attached, not as an improvement claim. Dollars not OEE points, a range not a single number, gates with thresholds and owners rather than a promise, and an explicit fallback that says when the line gets ordered and what the delay will have cost.
What should a capex-avoidance case include? The seven blocks above: decision, measured baseline, options, achievable value, sensitivities, gates, fallback. The numbers finance will test are the recoverable points and their bucket logic, the demand use of recovered capacity, the margin basis, the program cost including internal time, the ramp, and the cost of being wrong.
What if the recovery program fails? The gates are designed to find out early. A day-30 failure withdraws the case for the cost of the measurement. A day-90 failure orders the line one quarter later than it would otherwise have been ordered, with the delay cost stated in the paper, and with the line specified against a measured bottleneck rather than a forecast.
Blocks two and six, the measured baseline and the gated proof plan, are what separate a capex-avoidance case from an OEE slide, and neither can be written from the shift reports and the ERP. Six weeks on one line produces both, and a board paper built on them is one the CFO can audit and the board can approve as a test.
The fixed-scope pilot is the first forty-two days of the proof plan: one line, an industrial sensor and hub installed with your team, the machine-verified baseline on a stated basis, the loss Pareto with recoverable and structural buckets, the recovered capacity valued as production value, contribution margin and achievable margin using the plant’s own SKU economics, and the three interventions with owners for the day-60 gate. The fee is fixed and credited in full against a first-year subscription if you roll out. The pilot runs on Fabrico’s manufacturing performance platform (MES, OEE, CMMS & AI), which connects machine data, OEE and loss analysis, production scheduling, SKU-level output value and maintenance in one system, so the gates in the board paper are read from the same data as the baseline.
Request a demo or download the one-page board paper template to see the seven blocks laid out.
Login
Blog
