Blog / Metrics
Metrics · Guide

Manufacturing KPI Tree: Build One That Adds Up

OL By OEE Lab Editorial |Updated 2 November 2026 |5 min read
Short answer

A manufacturing KPI tree breaks one top goal, such as weekly good output or cost per unit, into the drivers that produce it. Each level must multiply or add up to the level above, so that a change in any driver gives a number at the top. In the example here, that shows two OEE fixes beating overtime on both output and cost.

A real tree can be calculated

Many KPI trees are only lists with lines drawn between them. "Quality" sits under "customer satisfaction", and no one can say what a one-point change would do.

Use this test: can you change a number at the bottom and calculate the top? If you cannot, you have an org chart for metrics and not a tree.

Example: a tree for weekly output

Example: a line runs 80 scheduled hours a week with an ideal rate of 3,000 units an hour. The goal is 170,000 good units a week, and the line is below it.

Good units/week: 155,040x Scheduled hours: 80x Ideal rate: 3,000 per hourx OEE: 64.6%Availability: 80%stops, changeoversx Performance: 85%slow running, short stopsx Quality: 95%scrap, rework
Example numbers. 80 x 3,000 x 0.646 = 155,040, and 0.80 x 0.85 x 0.95 = 0.646.

The gap to the goal is 170,000 - 155,040 = 14,960 units. The tree now lets you price each way of closing that gap, before you pick one.

Which lever moves the top most?

Change one driver at a time, and leave the others alone. That gives you a number for each option, in the same unit.

8 more hours: 80 to 88+15,504
Availability: 80% to 85%+9,690
Performance: 85% to 90%+9,120
Quality: 95% to 98%+4,896
Example numbers. Bars are scaled to the largest gain. Each lever is changed alone from the 155,040 starting point.

On output alone, overtime wins: it is the only single lever that closes the gap. Stop there and you will make the common mistake, because output is only one branch.

Add the cost branch

A second branch shows what each option costs per unit. The tree for cost is simple: cost per good unit = line cost / good units.

Example: the line costs 1,200 an hour to run, in any currency, and an overtime hour costs 1,800. Material is left out, since it does not change with the choice.

OptionGood unitsLine costCost per unitMeets 170,000?
Today155,04096,0000.62No
8 overtime hours170,544110,4000.65Yes
A 85% and P 90%174,42096,0000.55Yes
Example numbers. The third option also has a cost of its own, which you must add: engineering time, parts and the time it takes.

Now the decision looks different. Overtime reaches the goal next week but raises unit cost, while two OEE fixes reach it at a lower cost once they are done.

The two are not exclusive. A sensible plan may use overtime for six weeks while the availability and speed work gets done, then remove it.

How to build your tree

1
Pick one top goal
Choose the number the business needs, such as good units a week or cost per unit.
2
Write the equation
Find the factors that multiply or add to give it. If they do not, change the factors.
3
Go down until someone owns it
Stop at a number a named person can change, like stop minutes for one cause.
4
Fill it with real data
Use last month's numbers. Blank boxes tell you what you do not measure.
5
Test each lever
Change one input by a realistic amount, and compare what happens at the top.
One tree per goal. A second goal, such as delivery, gets its own tree.

The lower branches

Under OEE, the math turns into minutes, which is the job of the OEE loss tree. That page puts every minute of the shift in one box, so use it as the lower part of this tree.

Below that, use a downtime Pareto to split stops by cause. At the bottom sit the numbers a crew can change, like the shop floor KPIs.

Common shapes for other goals

Output is not the only top goal, so here are other equations you can start from. Each one can be calculated.

  • Cost per unit = (material + labor + overhead) / good units. Cost per unit shows how output affects it.
  • Capacity = available hours x ideal rate x OEE. Try it in the capacity calculator.
  • Labor productivity = good units / labor hours.
  • Rolled yield = yield at step 1 x step 2 x step 3.
  • Lead time = work in process / throughput, as shown in manufacturing lead time.

Worked versions of these sit on the KPI formulas sheet.

Mistakes to avoid

  • Branches that do not add up. If "morale" is on the tree, it is a useful thing to care about, but it does not belong in the math.
  • Too deep too soon. Three or four levels are usually enough to reach a named owner.
  • Treating gains as promised. The tree shows what a change would be worth, not how hard it is to get.
  • Ignoring links between branches. A faster line can raise rejects, so watch quality when you push speed.
  • Output nobody can sell. Extra units only become cash if they are needed, so check demand first.
  • Never updating it. Rerun the numbers monthly, since the best lever changes once you fix the first one.

Draw yours on one page

Draw the tree for your main line with last month's data on one page. Work out the top number from the bottom ones, and check it against what you shipped.

If it does not match, you have found a data problem worth solving first. If it does, take the top lever to your daily production meeting, and put its numbers on the KPI board.

See it live in Fabrico

A tree needs real numbers in the OEE branch. Fabrico, a light MES for OEE, takes stops, speed and counts live from your machines and shows OEE per line and shift, so the lower branches stay filled. See how Fabrico measures OEE.

What people ask next

What is a KPI tree in manufacturing?

It is a diagram that breaks a top goal, like good units or cost per unit, into drivers that multiply or add up to it. It shows how a change at the bottom moves the top. See the worked formulas on the KPI formulas sheet.

What is the difference between a KPI tree and an OEE loss tree?

A KPI tree starts from a business goal, and OEE is one branch. A loss tree starts from time and shows where each minute goes. Use the OEE loss tree as the part below OEE.

How many levels should a KPI tree have?

Enough to reach a number that a named person can change, which is often three or four. Going deeper usually gives you data with no owner. Stop where someone can act.

Where does OEE sit in a KPI tree?

It sits between the business goal and the shop floor. Output and capacity depend on hours x ideal rate x OEE, and OEE breaks into availability, performance and quality.

Can I build a KPI tree in Excel?

Yes, and it is the best way to start, since each box is a cell and the top is a formula. Change one input and read the result. Keep one sheet per goal and update it monthly.

See OEE per line and shift on your own machines

Fabrico is a light MES for OEE. It takes stops, speed and counts from PLCs or IoT sensors, with add-on AI cameras counting output where a machine has no usable signal, and operators tag the cause at the line.

Book a Fabrico demo

← More OEE Lab guides · The OEE loss tree →

OEE essentials: OEE calculator · What is OEE · Six big losses · Availability · TEEP · Cost of downtime · OEE benchmark · Free OEE tracker · Free OEE software · How to track downtime · How to implement OEE · Light MES vs full MES