Clay Tech

"clay-works make things real"

translated from clazytech.com

Don't polish a product that won't sell a million units

Development never goes smoothly.

Specifications change mid-development, deadlines close in with no fix in sight, molds need rework, patent royalty negotiations break down — all kinds of problems get in the way. For what it’s worth, I even have experience with “a factory fire burning up the parts.”

A lot of people involved have real commitment to getting a product shipped. So nobody gives up over just any problem or challenge. But if you actually stop and think calmly about it, you often find you’ve pushed things to a point where you have to ask, “did we really need to go this far?” I have plenty of memories of doing exactly that myself. So partly as a note to myself, let me lay out here how to apply the “brakes.”

Start with the easy case: a clear one-time expense. Mold rework, PCB rework, the cost of resubmitting for a regulatory test, losses from parts already ordered ahead that then had to be changed — these all fall into this category. You can see how big an impact these have simply by dividing the loss by the planned shipment volume. Say mold rework costs $7,600, and the product’s planned shipment volume is 10,000 units — that’s a $0.76 increase in cost per unit. In reality you’d also need to factor in the time lost to the mold fix, the resulting opportunity cost, and labor-month rates, but I’ll leave those out here to keep things simple. Suppose the problem that triggered the mold rework was that a faint fuzziness was visible on part of the product’s surface component. Was spending $7,600 to fix that actually justified? Polishing the finished part would surely have been cheaper than raising the cost by $0.76 per unit.

Another example: wireless performance requirements differ by country. Suppose a product passes Japan’s regulations but fails Europe’s at some point. If fixing that costs $15,000, is that a reasonable call? The right way to decide is based on sales forecasts for how many units will sell in Europe. Depending on the product’s overall cost structure, is it really worth selling in Europe if it means squeezing the profit margin that hard?

When you handle each fix and each extra expense this way — dividing by sales volume and adding it into cost — you sometimes suddenly notice an inconsistency. “There’s no benefit to spending any more on this. In fact it’s a net loss.”

That said, I have worked on products where this kind of thinking barely mattered. One was a product that sold a million units. The other was a flagship high-end product.

A product that sells a million units counts as a serious hit by any measure. It’s a stable product, it spawns a lineup of derivative variants, and you can expect it to keep contributing to revenue next year and the year after. So in the logic above, the “sales volume” you divide by can be assumed to be an extremely large number. In other words: whether we lose a few tens of thousands of dollars here matters far less than making sure we ship the product properly and carry that through to this year’s revenue, and next year’s and the year after’s.

In fact, while I was working on this product, I flew overseas on business trips as often as needed, and when issues came up in the part an Indian venture partner was responsible for, we flew them to Japan, put the joint Japan-India project team in a room together, and dug into the development work.

This was also the series where I committed the major blunder of having to rework the entire first production lot. But even then, it came through clearly to me — a junior engineer at the time — that the priority in a situation like that wasn’t cost, it was sticking to the schedule and getting the product to market. That mattered more than anything else. After all, even if you lose $76,000, when you divide that across the lifetime volume of all related models, the impact on cost per unit comes out to just a few cents. As an aside, I’ve heard that the first production lot of the PlayStation 2 apparently failed on unwanted radiation emissions, and shipped with radio wave absorbers and metal sheets plastered all over the inside of the enclosure. I have no trouble imagining that this, too, was the result of prioritizing the release date above everything else.

When I worked on the high-end flagship development, the development team was large. I think it was the biggest team I’ve ever worked with. In a large team, personnel costs usually make up a very large share of total spending. According to the National Tax Agency, the average salary for a regular employee in Japan is about $38,000, so using that figure, suppose you run a team of 100 people for a year. That alone comes to $3.8 million. But personnel costs aren’t just salary. There are various allowances and benefits, overhead for indirect departments that aren’t development roles, and the space people need to work along with the facility maintenance that comes with it — in practice, adding people brings all sorts of hidden costs behind the scenes. The common rule of thumb is that roughly tripling the number gets you in the right ballpark. That comes to $11.4 million.

A high-end flagship generally doesn’t sell in huge volumes. Instead, it keeps selling over many years — say, a total of 50,000 units sold over five years. Dividing $11.4 million by 50,000 units comes out to a startling $228 per unit. It’s a shocking number, but this is exactly the point where you need to stay calm instead of reacting hysterically. The question is: what price is this product actually going to be sold at?

Even within “high-end,” the going price varies enormously depending on the category. Within the range of electronics I’m familiar with, the areas where high-end pricing diverges most sharply from the general market rate are medical devices and luxury goods like audio equipment. Medical devices vary a lot by type and have a somewhat complicated cost structure, so let’s consider audio here since it’s easier to follow. Take CD players, for instance — you might think nobody uses them anymore in this day and age, but in enthusiast circles they’re still very much in use. And it’s common to find ones priced above $7,600. It’s hardly unusual to find a full audio system, speakers included, that costs more than $76,000 for a complete theater setup. Coming back to that $228 figure from before, without even running the numbers in detail you can tell this isn’t a big deal. In other words, here the question is how far you can push your obsession to build something that truly satisfies the customer. It doesn’t matter how many people it takes or how much it costs to get there (that’s a bit of an exaggeration, but you get the idea).

We’ve looked at two extreme cases, but both are more or less exceptions. For most product development, in general, you have to keep asking yourself, “is this really worth doing, to this extent?” There’s no end to how far you can push to improve performance. Wanting to meet users’ and stakeholders’ requests as much as possible is a perfectly legitimate feeling. And it’s entirely understandable to feel frustrated about shipping with even a small bug still left in.

But as long as you’re running product development and sales as a business, there are clear points where you have to strike a balance. If you want to push performance or quality further beyond that point, you need a properly reasoned answer ready for the question: “So you’re saying this will sell a million units?” “So you’re saying this will sell for $7,600?”

The content of this post is an excerpt (original text) from the following book. If you’re interested, please get a copy of the book.

The Shape of a Happy IoT Startup

The Shape of a Happy IoT Startup


Originally published in Japanese at https://clazytech.com/2022/08/1062/. Translated with LLM assistance and reviewed before publication.