Yoctopuce was created 10 years ago this month. So, this week, we offer you some thoughts on these 10 years that have gone by at a frantic pace.
Once upon a time Ten years ago, three employees from a world-renowned computer company made the discovery, more or less by chance, that by putting their individual knowledge in common they knew how to build USB electronic interfaces. The working atmosphere imposed by their management at the time was so deplorable that they decided to quit the "comfort" of a well-paid job to try their luck in creating their own company: Yoctopuce.
One must admit that it was a risky move since
- They only had a computer science background, their knowledge of electronics coming from working on personal projects
- Switzerland, a bastion of high prices, isn't the best place to produce electronics at a reasonable cost
The idea was to build USB interfaces such as those we would have loved to have when we were computer science students. It was clear from the start that we couldn't compete prices of products manufactured in Asia, so we thought that we could perhaps compensate by caring for quality, ease of use, documentation, and support. So, did we win our bet?
The light side of the force
Let's start with the positive and pleasant side of the experience.
We started small, with only 7 products: the Yocto-Knob, Yocto-Relay, Yocto-PowerRelay, Yocto-Meteo, Yocto-Temperature, Yocto-Color, and Yocto-Servo. Today, not only these products are still there, but a plethora of other modules have joined them: we have today about sixty different models in the catalog, without counting customized products.
We have always made it a point of honor to respect backwards compatibility as much as possible: each evolution of a product is usually 100% compatible with previous versions, both at the software level and at the mechanical level.
One of the great anxieties of the beginnings of Yoctopuce was the issue of product life expectancy: we didn't have the slightest idea. That's why we decided to systematically select the best quality that we could afford for the components. We guess it was a good choice because we can now affirm that as long as you take good care of it, a Yoctopuce module can easily work 24/7 for at least ten years.
We soon realized that selling naked electronic cards wasn't enough to satisfy the customers. You need to also sell accessories such as enclosures. We chose to have injected plastic enclosures made by a local manufacturer. We never regretted this decision, but the entry cost was really steep: it took us years to write off the cost of the mold.
If, at the beginning, we bought accessories such as USB cables and antennas in China from more or less anonymous resellers, we finally settled on European resellers. The costs are naturally higher, but the quality is much easier to check. As accessories are always sold for the lowest price that we can afford, price increases linked to suppliers changes have gone relatively unnoticed.
The design of modules is performed completely in-house. We decided very early on to focus everything on modularity. There are two module widths, the length is used as an adjustment variable. The USB electronics is always more or less the same, and therefore many electronic components are common to all products. So much so that we can now design a quite simple sensor in less than an afternoon and have a good chance for it to work the first time round.
We put our expertise as computer specialists to good use to automatize eventhing that could be automated. Whole sections of Yoctopuce manage on their own without us having anything to do other than check that it goes well. Production files are automatically generated, and the same applies for the documentations, the programming libraries, and a large part of the firmware. Our online shop lives its life without human intervention, the preparation of commands is supervised by a computer, the paperwork for international shipments is generated automatically... and there are many more examples.
Actually, there is a large software component in the Yoctopuce ecosystem: the web site, the automatic generators, the firmware, the programming libraries, and the applications available to the customers represent huge amounts of source code, so much so that we gave up trying to count the number of lines of code. Very early we decided not only to offer free of charge the Yoctopuce libraries and the few applications that come with them, but also to publish their source code. This seemed normal to us, notwithstanding all the work represented by this code. We realized later on that this is quite rare among hardware manufacturers.
The printed circuit boards used to build the Yoctopuce modules are ordered from a Swiss broker, who has them manufactured in Asia. Manufacturing PCBs requires machines and a know-how that we'll probably never have, but we are thinking about options to have these PCBs manufactured in Europe. When we have received the PCB, the remainder of the manufacturing is performed completely in-house: we have a serigraphy machine to deposit the solder paste, a auto-placer to deposit the components, and a reflow oven for SMD soldering. Two years ago, tired from soldering through-hole components manually, we even managed to buy a selective soldering machine.
Performing all the manufacturing in-house is a great advantage: not only do we have a total control over quality, but we can also produce in relatively small batches. If a product is about to be sold-out, we only need to turn the machines on and to run a production batch. We simply need to monitor the level of PCB stocks.
At the beginning, we thought that the Yoctopuce products would overjoy do-it-yourselfers like ourselves. Even if we have many individuals among our customers, there are not enough of them to feed us: our products are too expensive for them. On the other hand, professionals love us, for them we are rather cheap compared to proprietary automation solutions. Most of the time, we don't know what the customers do with our products, but from the little that we know, there are two main tendencies: building automatic systems to control quality at the end of production lines, and the embedding into commercial products.
Fun fact, we often see an addiction phenomenon with the customers: we regularly see a relatively modest order from a new customer, and a few weeks later, this same customer comes back to put a huge order with a bit of everything in it.
Thought originally as a way to be of service, the sale of custom products, which are only variants of existing products, now represent a non-negligible part of our income. We decided to offer this service without any of the additional costs which are usually expected: no administrative fees, no set-up costs, no design costs, and so on. For us, a custom product is a product like any other, the difference is that it's sold to only one customer and that there is a minimum quantity that the customer must order. It seems that this model, very advantageous for the customer, pays rather well on the long term.
Yoctopuce doesn't advertise. The first few years, we spent some money in Google campaigns, but we stopped after a while, apparently we don't need them anymore. We suppose that our web site has enough serious content to be credible for the different search engines. We suspect that word of mouth works also quite well: some large companies have research labs all over the planet and we see passion for Yoctopuce products spreading from one lab to the other without having been contacted by any of them.
Manufacturing quality products, ease of use accompanied by serious documentation has an interesting side effect. There is almost no support: customers usually manage on their own without our help. If we nevertheless notice that there is a recurring question, we write a post in the blog detailing the answer, which enables us to direct the next customers to this answer.
The dark side
However, don't imagine that everything is sweet and easy at Yoctopuce. If it was so easy to create one's own company, everyone would do it.
The main reproach for Yoctopuce products is their price. Believe it or not, Yoctopuce wasn't created to make us rich, we are not even sure that one can become rich while manufacturing hardware nowadays. The aim for Yoctopuce was simply to have an independent and motivating job, while having if possible happy customers. That's why we decided on a very simple pricing model: for each electronic product, we compute its cost in components and PCB, we multiply the result by three, and we obtain the selling price per unit. If you tend to believe that Yoctopuce modules grow on trees, that may seem a lot. But experience shows that this is a minimum: these two thirds of "profit" are used to pay salaries, buy and maintain machines, pay certification tests, buy consumable, pay insurances, pay shipping costs... the list seems almost endless.
If the first five years were touch-and-go budget-wise, we haven't had too many problems to make ends meet for a few years now.
We never talked about it on this blog, but our large customers know it: Yoctopuce started with three full-timers in a 50m2 garage, and a part-timer working from home. Ten years later, it's still three people in a garage transformed into a micro-factory, and a part-timer. We never dared to hire additional people without being able to guarantee a decent salary, an open-ended contract, and a sufficiently spacious workplace. Consequently, as sales increase, the work load increases as well. We believe that's where the true price of independence is: our work hours would make even the most liberal union rep. All of that for a decent salary, no more.
One could believe that it's possible to buy electronic components like you buy packets of potato chips. Nothing is less true, it's not uncommon to have lead time of several months. The issue was truly critical in 2016 when there was a shortage of MLCC capacitors. We ended up taking the problem head-on and we now have one year of stock for almost all our components. This represents huge sums of money tied up. But it's the only way we could think of to be sure to deliver any of our products within a reasonable time.
Yoctopuce sells over the internet and ships its ware (almost) anywhere in the world. Most of the time, everything goes quite well. But if there is a hitch, such as a damaged parcel, a tardy delivery, or even a loss, you can be sure that the sender will bear the costs. Apparently, in the world of freight transport, no one wants to assume responsibility. And that's a truly depressing issue because there is no solution as we are not going to create our own delivery company.
At any rate in Europe, one cannot manufacture and sell electronic circuits with total freedom, there are a number of legal rules to follow. These rules impact numerous domains such as safety, environmental impact, electromagnetic emissions, robustness to disturbances, and so on. It's rather normal that there are these rules, and respecting them tends to encourage people to build good products. What is more annoying is that the laws are never frozen, we even noticed that they change faster and faster. For example, the REACH laws change about every six months. Constantly adapting rapidly costs a lot of time and money, especially for small companies like our own.
After ten years, we can conclude that it is quite possible to manufacture electronics in Switzerland if you bank on quality rather than on selling price. Being in charge of your own company and having real decision-making power over everything that happens in it is far more interesting than feeling like an interchangeable cog in a company with hundreds of thousands of employees. However, one must expect many personal sacrifices. This being said, we don't regret anything and we certainly intend to keep up the same trend.