Parts Per Million

Parts per million is the language of quality at scale. A shop selling fasteners by the truckload, a contract manufacturer shipping consumer electronics components, a parts supplier feeding an OEM line cannot meaningfully talk about quality in percentage points. The interesting differences live in the third and fourth decimal place. PPM moves the decimal point to where the conversation can happen.

"Zero is the target; the number on the chart is whether you are getting there."

How parts per million works

PPM normalizes defect rates to a million parts, which makes small differences visible. A shop running at 0.04 percent defects and another running at 0.01 percent are both "very low defect" by percentage thinking. In PPM they are 400 and 100. Different worlds. The supplier scorecards your customers maintain run in PPM precisely so they can see those differences and rank you against the other shops competing for the same business.

Where the count gets honest or dishonest

The definition of "defect" makes or breaks the metric:

• Internal defects. Parts caught at your own final inspection and never shipped. These count internally for problem-solving but are not what the customer sees.
• Customer-reported defects. Parts the customer rejected after receipt. This is the PPM most customer scorecards report.
• Field defects. Parts that failed downstream in the customer's product. The most expensive category and the one most likely to cost you the relationship.

Best practice is to track all three separately and roll them up when needed. A shop with low customer-reported PPM but rising field defects is heading toward a recall conversation it does not see coming.

Where parts per million fits on the shop floor

Imagine a 40-person precision stamping shop running brackets for a tier-one HVAC supplier. The customer's scorecard sets a target of 250 PPM. The shop ran at 180 last quarter and management was pleased. Then a customer call: a batch shipped two weeks ago has a dimensional drift the customer caught in their own inspection. The drift is across about 400 parts in a shipment of 8,000. The PPM contribution from that one batch is 50,000 against the year's denominator and the quarterly average jumps from 180 to over 600.

The diagnosis is not interesting. A worn punch produced borderline parts. The interesting part is what happens next. The shop walks back from the customer-facing number to the internal data. Their own final inspection had flagged a slight increase in dimensional rejects two weeks earlier but no one had pulled the thread. A simple control chart on the daily rejects from final, posted on the floor next to the press, would have caught the drift before the customer did. PPM is the customer's grade. The internal data is the early warning. Both matter, and they should not look like the same metric.

Common mistakes with parts per million

• Counting only formal RMAs. Customers who quietly sort bad parts on their dock and never send them back are still unhappy. Their unhappiness will show up later as a lost contract. Talk to them about what they actually see.
• Averaging across products that have nothing in common. Stamped brackets and machined fittings have different defect profiles. A shop-wide PPM number averages them into noise.
• Treating PPM as a target to chase quarterly. PPM is a result of process control. Push for the number without fixing process drift and the next batch will undo the quarter's progress.
• Forgetting field defects. Customer-reported PPM looks clean while parts fail downstream in the customer's product. By the time you see the data, the customer has already started qualifying a second source.
• No defect taxonomy. "Defective" without a defect-mode breakdown is unactionable. The top three modes usually drive most of the number; you cannot find them without tracking them.

Parts per million and related Lean tools

Parts per million is a customer-facing cousin of defects per million opportunities, which Six Sigma uses internally to normalize for the number of failure chances in each part. It overlaps closely with defects per unit and is a stricter view of scrap rate. A shop with a clean first-pass yield almost always has low PPM, because the same process discipline produces both.