Now that I'm finally back at a real keyboard and have a minute...For me, sample size is a critical calculation. Each subject is expensive so you don't want to use more than you need. But, the cost of each trial is huge, so you don't want to fail by having too small a sample and have to repeat it.
Two sides of the same coin. For a typical manufacturing process, a minimum of 10 (or 10%) for n is required, but for companies in the food and health business, higher standards quality control is needed, meaning more precise statistics, meaning a higher minimum for n.
It's the same reason food companies use stainless steel everything in the factories, along with high-frequency start TIG welding. The most common, rather the cheaper, and the method most people first learn, is scratch-start. The problem with scratch start is you have to scratch the surface to start the weld... and there's a possibility for bacteria to grow in that scratch.
Essentially, different fields may have higher standards, but the lowest standard for a standard sample would be 10 (or 10% if you know the population of the thing being checked)