A bottling operation runs six lines. The OEE dashboard shows availability loss at 6%, performance at 5%, and quality at 3%. Nothing alarming. No red lights.
Translate those percentages to dollars, and the picture changes. Availability losses alone cost $39,000. Performance cost $30,000. Quality cost $19,000. The 14-point gap between the site’s 85% OEE target and the 71% actual OEE resulted in $88,000 in lost production value. One site, one day.
This is the pattern we keep seeing. Operations quotes OEE percentages. Finance thinks in dollars. They’re looking at the same operation and drawing different conclusions. The moment you put dollar figures next to the percentages, the conversation changes, because both sides are finally reading the same scoreboard.
This article shows how to build that translation. It covers the OEE waterfall decomposition (and why the naive method overstates losses by ~5%), introduces loss per OEE point as a metric for comparing sites, and works through a capital allocation example with real production data.
The OEE Waterfall: Why Losses Aren’t Additive
Most teams translate OEE losses to dollars by simply multiplying each loss percentage by daily production value. That’s wrong, and the error matters.
OEE is multiplicative: Availability × Performance × Quality. That means performance losses apply only to the time the line was available, and quality losses apply only to what the line actually produced. The naive approach (adding the percentages and multiplying by production value) double-counts because it ignores these dependencies.
The correct method is the waterfall decomposition:
| Loss Category | Formula | What It Means |
|---|---|---|
| Availability $ | A_loss × DPV | Cost of downtime against the full production day |
| Performance $ | A × P_loss × DPV | Cost of slow running, but only during available time |
| Quality $ | A × P × Q_loss × DPV | Cost of defects, but only on what was actually produced |
Where DPV is Daily Production Value at target OEE, A is availability rate (1 − A_loss), P is performance rate (1 − P_loss), and Q is quality rate (1 − Q_loss).
The key insight: each layer of the waterfall builds on the previous one. Performance losses are discounted by availability because you can’t lose speed during time you were already down. Quality losses are discounted by both because you can’t scrap units you never produced.
Worked Example
A bottling operation runs six lines. Combined baseline throughput: 255,000 units/day at $2.47 per unit. Daily production value: $630,000.
| Component | Loss % | Rate | Formula | Dollar Loss |
|---|---|---|---|---|
| Availability | 6.2% | A = 0.938 | 0.062 × $630,000 | $39,060 |
| Performance | 5.1% | P = 0.949 | 0.938 × 0.051 × $630,000 | $30,138 |
| Quality | 3.4% | Q = 0.966 | 0.938 × 0.949 × 0.034 × $630,000 | $19,067 |
| Total | $88,265 |
Compare that to the naive approach (simply adding the loss percentages):
| Method | Calculation | Total Loss | Error |
|---|---|---|---|
| Naive (additive) | (6.2% + 5.1% + 3.4%) × $630,000 | $92,610 | +$4,345 overstated |
| Waterfall | See above | $88,265 | Correct |
The 5% overstatement from the naive method might seem small. But it compounds across sites and days. Over a month, the naive method attributes roughly $130,000 in losses that don’t exist. More importantly, it distorts the relative ranking of loss categories. The naive method overstates quality losses (because it doesn’t discount for availability and performance) and understates availability losses as a share of total. That distortion sends capital to the wrong problem.
What the Waterfall Reveals
Look at the dollar column. Availability (6.2%) costs $39,000 and performance (5.1%) costs $30,000. On a percentage dashboard, those look like different urgencies. In dollars, they’re competing priorities for the same capital budget.
Quality (3.4%) costs $19,000, half the availability loss. The percentages (3.4% vs 6.2%) don’t reveal that 1:2 ratio. When operations, maintenance, and quality teams are fighting for budget, the waterfall decomposition turns a political argument into arithmetic.
Loss Per OEE Point: A Metric for Multi-Site Capital Decisions
The waterfall tells you where money is going within a site. But most operations run multiple sites, and the capital question becomes: which site gets the investment?
We propose a simple metric: loss per OEE point.
Loss Per Point = Daily Dollar Loss ÷ OEE Gap from Target
This normalises financial impact by the size of the improvement opportunity. A site with a large gap and high throughput will have a high loss per point. A site with a smaller gap or lower volume will have a low one.
Here’s how it looks across three sites running similar operations:
| Site | Actual OEE | Target OEE | Gap | Daily Loss | Loss per Point |
|---|---|---|---|---|---|
| A | 71% | 85% | 14 pts | $88,337 | $6,310/point |
| B | 73% | 93% | 20 pts | $22,410 | $1,120/point |
| C | 76% | 92% | 16 pts | $5,540 | $346/point |
Each site has its own OEE target based on equipment profile, product mix, and line age. That’s normal. The point isn’t to compare targets; it’s to compare the financial cost of each point of underperformance.
Site A’s loss per point ($6,310) is 18x Site C’s ($346). That difference has nothing to do with operational discipline or maintenance quality. It’s driven by volume and unit value. Site A processes more units at a higher value. On a per-unit basis, each percentage point of loss is materially more expensive.
If you ranked these sites by OEE percentage, Site C looks best and Site A looks worst, a 5-point gap. By daily dollar loss, Site A is losing 16x more than Site C. Loss per point tells you where each dollar of improvement spend will return the most.
Capital Allocation: Where the Waterfall Changes Decisions
Here’s the practical test. Your maintenance budget is $50,000. Site A and Site C both have availability problems. Where does the money go?
| Site A | Site C | |
|---|---|---|
| Availability gap from target | $39,000/day | $2,200/day |
| Investment | $50,000 | $50,000 |
| Payback (closing gap to target) | 1.3 days | 22.7 days |
Same investment, same type of fix, 18x difference in payback. The technical work is similar at both sites. The financial return is not.
I’ve watched companies allocate capital by chasing the worst percentage: “Site A is 71%, we need to get it to 80%.” That framing happens to reach the right answer here, but only by coincidence. If Site C had the lower OEE but also lower throughput, chasing the worst percentage would route capital to the smaller financial problem. Loss per point catches that. Percentages don’t.
How to Build This
You need four data streams to run the waterfall decomposition daily:
| Data Stream | Source | What It Provides |
|---|---|---|
| OEE decomposition | MES | Availability, Performance, Quality percentages |
| Actual throughput | Production system | Units produced per day |
| Unit economics | ERP / pricing | Revenue or cost per unit |
| Baseline comparison | Engineering / planning | Target OEE by site and line |
None of this is exotic. Most plants already capture it. The problem is that it lives in three or four systems, and most teams only stitch it together for monthly reviews. By then, the daily patterns are buried in averages and the money is already gone.
You can build this in a spreadsheet if the cadence is weekly or monthly. You can get further with a Power BI dashboard or custom SQL pulling from your MES and ERP, though you’re maintaining the integration yourself. A purpose-built operational intelligence platform does the stitching and the waterfall decomposition daily, without the manual plumbing, and puts the dollar figures next to the percentages so operations and finance are looking at the same screen.
However you build it, the waterfall formulas and the loss-per-point metric work the same way. The math doesn’t care what tool you use. What matters is that someone is running it, every day, at a cadence where the numbers are still traceable to specific events rather than lost in a monthly average.
Ready to translate your OEE gaps into dollar impact? Capstone calculates these financial bridges every day, across every line, once your data streams are connected. Schedule a demo to see how your operations look when measured in the language finance actually speaks.