[officialchicken]

I'm having trouble with this article and I hope someone from HN will help clarify for me: why was an initial run of 5000 units selected? This seemingly arbitrary number seems too important in HW development to be relegated to a single non-explanatory bullet point.

If my understanding is correct, product dev teams seem to aspire to an initial 500 - 2000 units for that initial run as the symbolic point of reaching this "finalized" level of development. Any production at this level is considered a finished good ready for the sales/supplier pipeline.

Once a product has reached this level, my understanding is that subsequent runs is typically for 10K to 25K widgets per run ("lot") that see anywhere from 10,000 to 250,000+ widgets for a single SKU per production run.

Above 1 million units, everything changes, but the 5K first-and-only run seems abnormal.

[smarinov]

Producing 500 or 2000 units would virtually require the same time, so you may as well get a standard batch size and improve your margins.

[officialchicken]

There's a big difference between 2000 and 5000 unit initial runs... why 5000?

[Gibbon1]

For an initial production run there are a lot of fixed setup costs. Just from my gut feel 5000 units is just barely off the NRE (non recurring engineering cost) 'knee'. (And the manufacturing learning cost knee) Less than 5000 units the NRE costs start to dominate your out the door manufacturing price. So at 2000 units that might be 50% to 80% higher per unit than at 5000 units.

Also people often forget about your suppliers management, accounting, and sales overhead. That means that they will have minimum orders below which they don't make enough money justify the overhead of shepherding your product through their shop. For Chinese suppliers 5000 units appears to be typical because lower volumes are typically on-shored. On shoring is going to increase your costs by probably 30%.

You might ask the other question, why not 10,000 or 20,000 units? Answer is that starts to be a huge chunk of money for something that isn't proven.

So ans: 5000 because less the NRE kills you. And you'll have to on shore production. And no more than that because otherwise your initial production costs becomes the dominant part of the investment risk.

[smarinov]

Thanks for complementing my answer. That's exactly what I had in mind, but I (still) can't type from mobile devices. :)