It seems like a month can not go by without some large scale food recall, with the latest being approximately 31 million pounds of peanut butter (due to salmonella contamination). I started to wonder what the sigma level for the aforementioned peanut butter would look like. According to a major news website, at least 500 people had been affected by the contamination, definitely an error when listening to VoC. Assuming 500 illnesses with a batch size of 31 million pounds, I get a sigma value of 5.66. Of course this assumes batch size is the right opportunity to count.

When I teach Six Sigma, there is a slide explaining how sigma values are calculated followed by another slide with sample values for given industries. An example of an industry with a 6 sigma plus rating is airline safety. Most industries will operate around a 2-3 sigma value for a given process. An example of a process that typically operates around 3 sigma is the accuracy of restaurant bills. In other words, the process is 93.3% accurate or for every one million bills, or 66,800 will have an error.

So as a consumer, should I be happy with a food that currently has a value of 5.66 sigma? Is the media overreacting? I decided to do some research. I googled DPMO and food and after 14,000 hits, decided to narrow my search to the FDA website. The first thing I saw was a listing of over one hundred foods affected by the recent recall. Now if I assume a batch size per product or per manufacturer my number of opportunities just decreased significantly, however without a firm number of defects provided by the manufacturer, it would be impossible for me to calculate DPMO on a given brand name product.

I found my search proved better when I changed from DPMO to PPM (parts per million). It turns out the FDA has an entire handbook on what number of defects are acceptable for all foods. For example, the FDA considers a maximum defect level of 2% or more for apricots harvested for canning that have been damaged or infected by insects as acceptable. A 98.2% defect fee yield translates into a sigma value of around 3.6.Okay, so the process is better than restaurant bill accuracy but I’m not sure as a consumer I’m happy with this value (good thing I don’t eat apricots).

Looking at coffee beans, if a batch is less than 10% insect infested/damaged, it is considered acceptable. This translates into a sigma value of around 2.8. So I have a higher quality level with baggage handling at the airport (around 4 sigma) than purchasing green coffee beans. As someone who’s had their bags incorrectly routed twice, the results are unsettling.

In addition, the FDA website lists multiple types of defects (mold, insects, contamination, etc.). This gives one the availability to calculate DPU for some foods. After viewing the website (, I’m not any more worried about the peanut butter outbreak compared to any of the other foods I eat. Product traceability has allowed companies to pinpoint the exact extent of risk involved with the peanut butter and recall any questionable product. Whereas with other foods you purchase, the threat for defect is still there; it just isn’t as widely publicised by the media.

About the Author