DPMO: defects per million opportunities explained

The formula has three inputs:

DPMO = defects ÷ (units × opportunities per unit) × 1,000,000. The part in brackets is the total number of opportunities. Some teams first work out defects per opportunity (DPO), defects ÷ total opportunities, and multiply that by one million.

1. 1

   Define the unit. One part, one order, one invoice or one claim.

2. 2

   List the opportunities. Write down each requirement that could fail on a unit, tied to what the customer cares about.

3. 3

   Inspect and count units. Record how many units were checked in the period.

4. 4

   Count every defect. If one unit has three defects, count three.

5. 5

   Work out total opportunities. Units × opportunities per unit.

6. 6

   Apply the formula. Defects ÷ total opportunities × 1,000,000.

7. 7

   Compare. Against last period, a target or the customer's requirement.

These examples are illustrative. Notice how the opportunity count changes the picture:

Twenty defects in 200 assemblies sounds worse than 25 in 1,000 parts. Once opportunities are counted, the assembly line has the lower defect rate per chance. A DPMO of 5,000 does not mean 5,000 parts are defective; it means 5,000 defects per million chances.

Six Sigma tables convert DPMO to a sigma level. By convention they include a 1.5 sigma shift, which allows for a process drifting over the long term:

Yield here means the share of opportunities without a defect. The Six Sigma calculator works out the sigma level for any DPMO and lets you turn the shift off to see the long-term figure.

Each measure answers a different question, and they work best side by side:

Read them together. If DPMO rises while first pass yield holds steady, minor defects may be creeping up on units that still pass. If first pass yield falls while DPMO holds, defects may be bunching on fewer units or one step.

DPMO is only as good as the definitions behind it. Agree these before you report it:

• One unit definition: Used the same way by everyone.
• A written opportunity list: For each unit or process type, tied to customer or specification requirements.
• Objective defect criteria: So two inspectors classify the same condition the same way.
• A set period or sample: Decided in advance, so results are not picked after the fact.
• A rule for several defects on one unit: Usually count each one.
• A review when things change: New designs, workflows or requirements change the opportunity list.

Opportunities should be real chances for a defect that matters. Counting trivial ones, such as every screw on a cover, inflates the denominator and makes DPMO look better while nothing improves.

DPMO measures frequency. It leaves several questions open:

Late defects are the expensive ones. A 2002 NIST study estimated that inadequate software testing infrastructure cost the US economy $59.5 billion a year, much of it from defects found late or by users.

A single plant-wide DPMO hides more than it shows. Useful breakdowns:

1. 1

   Review the trend. Look for spikes, steady climbs and repeating patterns.

2. 2

   Break it down. Sort by defect type, product, line, shift or supplier.

3. 3

   Pick the biggest or most serious contributor. Weigh frequency and severity.

4. 4

   Check the measurement. Make sure everyone counts defects the same way.

5. 5

   Find the root cause. Use structured problem solving, often a DMAIC project.

6. 6

   Change the process. Update controls, settings, training or supplier requirements.

7. 7

   Watch DPMO afterward. A drop that does not last means the fix is not built into the process.