In this episode I speak with Dave Millman of SalesDev.Global about some of the various aspects of bringing a new hardware product to market that are most commonly underestimated by hardware startups.
Dave has decades of experience helping hardware startups and he specializes in helping them find their first customers. He is also one of the experts available to help you inside the Hardware Academy and he has even taught a couple of live workshops.
Podcast MP3: Download
Dave Millman of BizDev Global
John Teel: Today I’ve got Dave Millman on the podcast again from SalesDev.Global. Dave is an expert at helping hardware startups find their first customers.
He was actually my guest on episode number two, where we discussed the five biggest customer mistakes that hardware startups make, and then he was also a guest on episode number 12, where we talked about various horror stories of product development.
Today I’ve invited Dave back onto the podcast to discuss some of the areas that entrepreneurs most commonly underestimate. Welcome to the show, Dave.
Dave Millman: Great to be here again, John.
John: Thanks for coming back. You’re, I think, the first person I’ve had back three times, so congratulations on that.
Dave: I would say you could take me to dinner except nobody takes anybody to dinner anymore.
John: Yes. [laughs] I’ll send you a gift certificate.
Dave: There we go.
John: Today we were going to talk about various things that hardware startups tend to most commonly underestimate. Let’s jump right in. Dave, I’ll let you introduce the first point and then we’ll just go from there.
Dave: Great. John, isn’t it remarkable how when you’re talking to somebody who has never done hardware before, that they’re assuming that their very first product is going to be perfect and they’re going to get the electronics right the first time?
John: And they’re going to do it for a few hundred dollars or a few thousand dollars.
Dave: [laughs] As far as what people underestimate, the reality is that I think experienced engineers schedule multiple PCB revisions, and actually multiple revisions of almost everything, but let’s talk about PCBs first.
John: Yes, absolutely. I think the key point to keep in mind is that nothing is ever right the first time, as it’s extremely rare that the first version is going to be ready for production. That’s true in regards to any element of product development, but also, of course, with the printed circuit board.
Dave: You need to change the way the product works, sometimes you need to change the components on the product. There’s any number of reasons that you’re going to revise a PCB.
John: Absolutely. I don’t think I’ve ever seen any design that had it right the first time. No matter how experienced you are there’s always little tweaks that are going to be required. If you can get that first version back and it’s fully functional and it only needs a few hardware tweaks, then I think that’s a pretty good job.
Dave: The most common tweaks that I’ve seen be required is the onboard antenna is wrong or is blocked, or needs to be moved, or the mounting holes are wrong because of revisions being made to the enclosure, or the new holes need to be drilled to mount a different jack or a different plug or a different switch or a different power supply, or things like that. Those seem to happen on virtually every design.
John: Absolutely. I’ve seen very similar things. In general, the more complex the product, I think the more iterations that are going to be required.
That’s one of the many reasons I’ve really tried to emphasize how important it is to simplify your product as much as possible upfront because that’s just going to lower how many iterations of your printed circuit board or all of your prototypes are required, which is going to translate into faster to market and less development costs.
Dave: If anybody has any doubt that what we’re saying is true, just open up an electronic device and look at the PCB. Look on the silkscreen, which is the lettering on the PCB, and find where it says rev and see how often that rev is something other than 1 or A.
John: That’s a great point. If you can find one that has rev 1 or rev A, that would be pretty rare I think.
Dave: In my case, the stove that burned out the controller board about a year, a year and a half ago, it was in fact a rev A and it burned out after 18 months. When I got the new one, it was rev H if I recall, or something like that.
John: The rev As do make it out there but they obviously can have some serious issues like that.
Dave: It was remarkable actually, how many components had been replaced.
John: A lot can happen between A and H, I think. Just in general, simplify the design as much as possible is probably your best bet at trying to reduce the number of iterations, and using things like modules.
Like you mentioned, antennas are always notoriously hard to get right. Anything wireless, especially if it’s custom-designed, is always going to be problematic. I found anything with GPS tends to take a couple iterations to get just right if you’re using a custom antenna, mainly because the signals coming from the satellites are so weak that you really have to have everything perfectly tuned.
Then complex, like microprocessor designs, if you’re trying to design that from scratch, that’s another type of circuit that I find tends to require multiple iterations to get right.
Dave: Well, most designs are small these days. Most products are small these days. Anything mobile or wireless tends to be small. As you’re tweaking the design and making it even smaller, inevitably the mounting holes change.
John: Yes. It’s kind of what seems like a small detail but you’re right, that’s one I’ve seen burn a lot of people.
Dave: Yes, and if you got to move a mounting hole then there’s a really good chance you’re going to have to move some traces and make it a component.
John: Or some components, yes. Like you said, most products are small so everything is packed really tight and that just gives you a lot less flexibility when any type of change is required, whether it’s a mounting hole or any other type of change to the board. When everything is packed tight, you make one little change, now you have to move everything out of the way. That’s why things where size is supercritical like wearables, that tends to add, I think, an additional level of complexity.
Dave: Which really guides us to the next topic here. We talked about design changes, but just to hit our second point, most products require updates at some point in their life, right?
John: Absolutely. Even though you may get the product right for that current version, to get that released over the life of the product you’re going to have to do most likely several updates on that product.
Dave: Let’s run through those reasons. If you’re planning an initial product run of 1,000 units or 5,000 units, you’ll make certain decisions based on availability of parts and cost of parts and everything else.
But if the product is successful and you start to plan a production run of 20,000 or 50,000 or 100,000 units or more, the costs of the same parts are going to be typically changed a lot, and the more complex parts are going to be the ones that change where their price tends to drop the most, right?
Dave: That’s where redesigns come in. All of a sudden, gee, at 100,000 units I can afford to go to this other processor or this other chip that has these other features. That will allow me to drop this other chip that I had on there, and instantly we’re doing a design revision.
John: Yes. Similar with a module. A lot of designs may start off with a module to simplify development, but once they’re in massive production producing hundreds of thousands, then it can be beneficial to replace that module with a custom circuit to get your cost down and increase your profit margin. That’s almost always going to be updates are going to be required at various points in the product cycle or product lifetime constantly trying to get that cost down.
Dave: When you say modules, you’re talking about Bluetooth modules, wireless modules?
John: Exactly. Microprocessor modules and wireless modules are the two most common types of modules.
Dave: Then, frankly, products don’t stand still. How often does a product stay in production without changes for more than, let’s just say, two or three years?
John: Yes, I think that’s pretty rare.
Dave: Probably 10% or 20% of products or less stay in production without changes.
John: Yes. In life, I think we can all agree nothing’s ever static, everything’s always changing. That’s just going to be the same way with a product.
Dave: I remember I was evaluating a PlayStation 3. PlayStation 4 is an established product already and PlayStation 5 is coming up for announcement here, but when I was looking at a PlayStation 3 a few years ago, they had just introduced the first of the cost reduction redesigns.
I had a choice of buying one that was compatible with PlayStation 2 and another one that was cheaper and smaller and consumed less power and fit in the cabinet better, but it wasn’t compatible with PlayStation 2. They had designed that feature out as a cost reduction and that was a change, but it was still a PlayStation 3, right?
John: Yes, that’s a great example. The first version had to be backward compatible, but after the product’s been out for a while then they didn’t need that anymore. Design that out, that reduced the cost. I think that’s a great example.
Dave: Our listeners here would probably love to be as successful as the PlayStation 3, right?
John: Yes. [chuckles]
Dave: But these things happen. Three years into the product and all of a sudden the four-channel chip is available for less than you were paying originally for the two-channel chip. The four-channel chip allows you to do oh, so much more, right?
John: Yes, or just components go out of service. There may be something you designed in early on and it was going to be available for a couple of years, but after a couple of years that manufacturer’s phased out that component and now you have to do an update to your design to use their new chip or to find an alternative solution.
Dave: Oh my gosh. All the wearables guys go through this because it seems like the displays never stay in production.
John: Yes, that’s a good one.
Dave: Just building on what you said, the displays are typically available for a year or two and that’s it, they’re on to the next display. There’s lower power and more colors, and nobody’s going to buy the old one anymore anyway, so why would they keep it in production?
John: Yes, that’s a good point. I can see that being constantly a changing target.
John: Is there anything else you want to cover on requiring the updates?
Dave: I don’t know, I guess it’s a guideline. If you’re designing a product that’s going to be connected to the cloud, then I think you probably need to design that product to be updatable from the cloud. I just switched gears slightly from hardware updates to firmware updates, but updates are updates.
So much is changing in cloud connectivity and cloud functionality these days. Two years from now, goodness knows what you’re going to want to connect your product to, or what enhancement you’re going to want to add to it. Right?
John: Yes. You definitely want the design to be able to have the ability to upgrade your firmware, that’s for sure. You want to go ahead and jump to number three?
Dave: You bet. The people who are new to engineering typically miss this one. All the experienced engineers listening are going to get this, but it seems like the folks who are new to hardware don’t recognize the need for prototyping, particularly prototyping on different hardware. Like prototyping on development kits or evaluation boards or FPGAs or whatever.
John: Yes, that’s something I’ve seen as well. They tend to want to just skip that step and go right for the full production version.
Dave: Right. They feel like, “Oh my God. Why would I design it twice, once on the dev kit and then once on my own custom version of it? That’s going to cost way more money and take way more time.”
I think the response from an experienced engineer would be, “Actually, it’s going to cost less money and less time because we’re going to work out all of the functionality and many of the bugs before we go to your custom board and your custom software and everything else, and avoid a much longer and more difficult debugging process.”
John: Yes, absolutely. The only time where I think it can make sense is if you happen to be a big company and you’ve got millions of dollars to spend, then you can get the product to market faster, most likely, if you dump all that money into starting with the custom design. A lot of billion-dollar tech companies, they may skip over the development kit prototype because it’s faster to just start with the custom design, but they have millions of dollars to spend.
I think most people listening to this podcast don’t have that kind of money to spend. In that case, for startups and entrepreneurs, it’s definitely something you do not want to skip over.
It’s so much more affordable to learn your lessons and to learn things about your product from a development kit prototype versus learning those after you’ve created the custom design. Otherwise, you’ll probably end up finding yourself having to do multiple custom designs instead of learning all those lessons from the development kit and then implementing those lessons learned into your first production design.
Dave: If you visit any of the established chip companies these days, TI, Analog Devices are really common, and you start looking at a power converter or a processor, whatever, the dev kit is discussed right on the page along with the part.
Because the dev kit is right at the top of the list that all the experienced engineers want to get ahold of as soon as possible to make their final decision about that part and about implementing it in their design.
John: Yes, absolutely. One point since you mentioned that is that a lot of entrepreneurs, they want to just immediately gravitate to an Arduino or Raspberry Pi. It’s a lot better if you can go with a development kit that’s close to what you want to have in the final version.
If you just have a simple product that can be best served with a microcontroller, you don’t want to design your proof of concept using a microprocessor-based system like a Raspberry Pi, because then to go from that development kit proof of concept prototype to the production version is going to be just almost like starting from scratch.
Try to find a development kit that uses the production version of what you eventually plan to use in the production version. For instance, any of the microcontroller manufacturers, they supply development kits for, say it’s an STM32.
You can get a development kit that’s based on that microcontroller and use that for your proof of concept instead of going with an Arduino and then migrating from there to something that’s not at all related to the Arduino.
Dave: Time to market is so critical these days that the big manufacturers are shipping development kits for power converter chips. That’s not a microcontroller or a processor, it’s just a 8-pin or 12-pin power converter, but there’s a prototyping board or a development kit available for it, because that lets their customers get to market quicker and provides higher value to their customers so that their customers buy their parts.
John: Yes, get to market quicker and cheaper. It’s obviously much– You’re saving money on development cost if you’re using a development kit versus custom designing that.
Power solutions are a good example with kids, typically if it’s a linear system that’s pretty simple, but for switching power supplies, it makes sense to use a module solution or a development kit, because it’s all about- you want to get that product out to market as fast and as cheap as possible so you can begin getting feedback from the market.
That’s always needs to be the priority is, quick and cheap to get that feedback. These are just some of the ways that– Development kits allow you to get that, to make that happen quicker.
Dave: That feedback is going to come from your engineering team, who’s going to say, “Wow, this part’s way more sensitive than we thought,” or that feedback is going to come from potential customers who’s going to say, “Wow, we love this product,” or, “Oh, you missed this feature.” That feedback you’re talking about is going to come from all over the place.
John: Absolutely. Yes. Good point, you get it from both engineering and from potential customers. I guess the one area, the one type of product that I find that is the most problematic to deal with development kits are wearables just because of the space requirements, that you can still do a proof of concept or really small development kits out there that make it a little more practical, but it’s always going to be somewhat restrictive in what you can do if you’re trying to make a smart watch out of development kits.
Dave: Yes. That’s a hard one.
John: Yes, and wearables, they tend to be extra tough in general because they’re harder to use development kits, and they require everything in the design be packed tighter and that just packing things tighter takes longer to do, and as I already mentioned, if you have to make any board changes, then it makes that even more problematic to try to squeeze in a change.
Dave: John, a very similar note to your wearables point is Magic Leap, the VR headset makers in Florida. They designed an amazing VR headset experience.
When they went to reduce the size of that to something that was wearable and under four ounces, or five ounces, whatever the weight limit was, they weren’t able to do it. If they had prototyped it faster and sooner, they could have learned that sooner and not set everybody’s expectations that set them up for a big fall recently at the beginning of 2020.
John: That’s why, even though it’s good to start with a development kit, you also want to be looking at the production version. The development kit version may not be the size that you need eventually, but you need to make sure upfront that the production version can hit that size.
You have to look at both the development kit, but your long-term view has to be on the production version. You don’t want to just proceed with the development kit, get through all that, and then start your production version before you realize that you’ll never be able to meet the size that you’re trying to do.
Dave: Realize that it’s going to weigh six kilograms and you need the size of a grapefruit.
John: Absolutely. Development kits are useful, you can get so much value out of them, you’re going to learn so much, it’s going to save you money, so don’t skip that step unless perhaps if you have a specific wearable product, then it may not always be feasible. Unless you have unlimited amounts of money, then, of course, why are you doing this if you’re already rich?
Dave: [laughs] I think we’ve hit the dev kit point clearly enough.
John: Yes, let’s move on to number four.
Dave: Oh my god, the elephant in the room. Enclosures. John, have you ever heard of a company, even an experienced company, getting the enclosure right on the first iteration?
John: Oh my gosh, definitely not the enclosure right. I feel like that, my experience that takes more iterations to get right than electronics does.
Dave: Today, the established production method for most high-value products is injection molding, and the things that are difficult to get right with injection molding are the appearance, which includes the colors, especially if you have different materials, matching the colors is always difficult, the finishes, the move from different types of molds to other types of molds is going to affect the finish, the fit, and when I say the fit, we’re talking about how two parts fit together.
The worst type of fit problem is the snaps and other opening and closing enclosures to bond together two halves of an enclosure. That’s the fit, the snap fits, the lids and things like that.
Then, the worst of all is the display screens, anything transparent or translucent, you’re going to be wrestling with the finish of it, the transparency of it, the thickness of it, and you just need to assume you’re going to be doing 10 iterations on that and be pleasantly surprised when it only takes you eight.
John: Yes, yes. A lot of these can really draw out the number of iterations, especially if appearance is really critical for your product that’s so hard you can’t always get a good feel for the appearance when it’s in a 3D model, so you have to do a lot of prototype iterations to get it to look exactly like you want.
Appearance is also something that requires so much back and forth with the industrial design or 3D modeler, because it’s hard to convey how you want something to look. I find appearance can really lengthen the time to get the enclosure ride and increase the number of iterations required to get it right.
Dave: Then, you have the perfect colors, but then you change one of the plastics to a stronger plastic or the base or something, and now the color’s wrong because the color in ABS is different from the color in another plastic.
John: Absolutely. Then, snap fits, you mentioned that, that’s another one that seems so simple to have two things snap together. Snap fits can be so problematic and just require a lot of iterations to get them right. Obviously, with modern 3D modeling software, you can get it pretty close, and especially if you have a lot of experience with snaps, and you know how much they have to overlap each other.
In general, snaps tend to take a lot of time to just get right. I know for my own product, which tended to be more mechanical than electrical, and it had a lot of snaps, it was like a snap that a customer- every time they use the product, the snap was engaged, or disengaged. It took me at least 10 prototype iterations, I think more closer to probably about a dozen to get that right, not just the snap, but all the other elements of the product.
I don’t ever find usually their printed circuit boards take that many iterations to get right unless you’re developing, I don’t know, trying to land your product on the moon or something really complicated like that, but in general, I find that the enclosures can take a lot of iterations and a lot of time to get right.
Dave: Of course, a whole new dimension of complexity comes in when you try to do overmolding.
John: Yes, if you require overmolding, which is when you have parts of two different materials embedded in the same part, like a plastic part that’s got some rubber sections of that plastic part. That’s what overmolding today is referring to.
Dave: It’s super common these days in handheld devices and cooking implements and just a whole wide range of applications. You’ve all got overmolded products. Well, now you’ve got two different plastics being molded together and that tends to, would you say it’s fair to say, double your number of iterations to get it right?
John: Yes, it’s probably close. I’ve not done a lot with overmolding, so I don’t really have a lot of data points to say for sure that it doubles it, but it definitely adds a lot of complexity.
Dave: Those are usually very high volume products anyway.
John: Yes. Like I said, it’s not something I dealt with a lot as overmolding, but it definitely adds a lot of complexity to the product.
Dave: When you take a brand new product out of a box and you notice that the overmolded soft rubber is already peeling out of one corner of it that you know they didn’t do enough iterations to get back.
John: Yes, exactly, exactly. Another thing that I have found that requires a lot of iterations on the prototype is, obviously when you’re doing your first prototypes, you’re probably using 3D printing, but 3D printing is not at all like the final production plastic.
You may think you’ve got your snap fits to be optimized for your 3D printed version, but once you transfer over to your production plastic, which has totally different characteristics, stiffness and things like that, then you can find that there’s still a lot of iterations required to get that just right.
That’s one more area where I found CNC machining because that allows you to start with a production plastic and carve out your product, so that can be used as helpful for getting snap fits and things that are optimized that are going to be dependent on the type of plastic.
Dave: For our mechanical friends, mechanical engineering friends out there, the stress profile of a 3D object that’s been 3D printed is as different as you can imagine, from the stress profile of a 3D object that has been injection molded.
John: Yes, I mean, one is, basically, layers stacked on top of each other, and the other one is a complete molded plastic part, so, they’re quite different.
Dave: Since we’re talking about snap fits and everything else, much depends on the skill of the mold makers, right? Because the most skilled mold makers will add fillets, will strengthen different areas where required, because they can visualize where the stress points are.
John: Yes, absolutely. Just another point, it’s always try to get all these iterations done before you get to the molding stage, because making changes to a mold is really complicated.
You can modify an existing mold by taking away metal, and then that adds plastic to your part in that area, but you can’t go the other way around. You can’t remove parts of your product, because that requires adding metal to your mold, and you can’t really add metal back to the mold. That’s like having your hair cut too short and then trying to glue hair back on.
I’m only speaking from personal experience recently, but having to cut my own hair due to the pandemic.
Dave: Every once in a while you will actually see a plastic part that has a odd-shaped, flattened surface to it in the back bottom somewhere, and that’s sometimes where a mold has been welded or repaired, or something.
John: Yes. Try to get all these iterations done before you go to the molding. Most likely, you’re going to still require at least one tweak to the mold, maybe optimize a finish or a snap fit or something like that. You want to try to have all the major issues resolved definitely before you get to the injection molding stage.
Dave: We’ve really already bled over into the next topic, which is- which of these typically cost more than expected. I think you just did a real good job covering the molds that cost more than expected.
John: Yes. I think two general statements you can make is everything costs more than you expect, and as we’ll get to for the next point, and it takes longer than you expect. That’s, I think, just a two general points you can make about any engineering product, whether it’s NASA or SpaceX, or you building a Bluetooth beacon, everything always is going to cost more and take longer than you expect.
I find that cost is one area that entrepreneurs definitely underestimate a lot. Sort of extreme examples, entrepreneurs reach out to me that have tried to hire someone through a website like Fiverr for as little as a few hundred dollars to design their printed circuit board, or even up to a few couple thousand dollars, but in every one of those cases, I’ve never seen them deliver a quality product.
Usually, it just ends up being money wasted or them having to try to get their money back on what they paid for. You have to have realistic expectations on the cost, otherwise– This is where you opened yourself up for being scammed or ripped off if you don’t have realistic expectations.
If you’re wanting to build, I don’t know, a new smartphone and an engineer tells you they can do it for $3,000, you need to run away from that engineer. Having realistic cost expectations can help you know when you’re dealing with an engineer that’s been truthful and realistic with you, or one that’s just trying to sugarcoat things, or tell you what they think you want to hear because they want your project.
Dave: John, let’s run down the list of all the things that tend to cause people to underestimate their costs. Let’s just hit the whole list of mold cost-kickers, the things that boost the cost of molds.
John: With molds, I would say the physical size of the product. Really large molds are really expensive, takes a lot of metal to make them and a lot of machining to remove that metal from a large mold, so they’re going to have a high cost because of the production time to machine that mold. Also, the number of cavities. The cheapest mold is going to be a single cavity, which means it produces one part with each shot of plastic. As you get higher production volumes, you want to add multiple cavities. That will lower your product cost, but it’s going to increase your mold costs so you want to start with one cavity and increase from there.
The hardness of the mold is also a big impact on the cost. If you start with, say, either a soft steel mold or an aluminum mold, that’s much easier for them to machine to make the mold, so it’s going to be a lot cheaper than if you use really hard steel.
Once you’re in a really high production volumes, and you want to use really hard metals, because you don’t want the mold to wear out, then the other thing that really impacts that are side actions.
I’m not going to go into too much detail on side actions because it’s hard to explain without some visuals, but side actions are basically moving parts of the mold that have to come in and out from the sides, and they just add a lot of complexity to the mold.
Those are the four things that I see that have the most impact on what the mold cost. Then, of course, the number of products you have, if your quote, one product really requires four different devices and each of those has their own enclosure, then now you have four times as many mold, so you have four times as much mold cost.
Then, you can always try to simplify or reduce the number of custom plastic pieces required for your product. If you can get it down from needing five plastic parts to only four, then that’s going to save you the cost of one mold.
Dave: Really good example of that that we’ve seen a lot of is an entrepreneur will have a great idea for a wireless device, a rechargeable device and they’ll have great ideas for a smart charger, not realizing that the smart charger is actually a separate product with separate PCB and separate enclosure.
John: Separate certifications and separate manufacturing setup, its everything is double. That’s probably one of the best ways to simplify your product. If your product is requiring more than one device, then try your best to get that down to one single device. One way to do that, let’s say, for it, I’ve seen a lot of product ideas where they have a core device that’s got wireless communication and then they have a display, a separate display device, so they want to have their own custom display.
No, you don’t want to do that. Instead, get rid of that custom display and replace it with an app on their smartphone or their tablet already. That’s just an easy way of getting rid of one of the two devices or like the example, you had mentioned is academy member who had a fairly simple device except he wanted a custom wireless charger that had motors and everything that was going to really add all this extra complexity.
It just made it so much simpler just to get rid of that wireless charging function and just replace it with more of a standard type of connector. That was one way to really reduce the cost and the complexity.
Dave: Then there’s the things that people don’t think about unless they’ve been through the process before. For example, documentation.
John: Yes, that’s always overlooked.
Dave: Some jurisdictions like Europe have very specific requirements about what needs to be in the documentation and the languages that the documentation needs to be in. In Canada, it has to be in French and English with equal size text for both on the outside of the box and documentation, if you’re intending to sell into a whole bunch of different countries, can add a significant cost to you.
John: Oh, absolutely. That’s one good reason to start with one country and then expand from there once you get that first one right. It’s like if you try to sell to the US and Europe and Asia, it’s just you’re creating so many nightmare situations for you. This is going to be overwhelming.
Dave: Of course, what needs to go on your documentation, but your certification and all your symbols, the UL and FCC and CE and RoHS, so all the certs. There are requirements about where those are shown on the package on the documentation and you got to get all that right, you got to get those certifications. There’s a lot of cost involved there.
John: Absolutely. You mentioned the BLE or Bluetooth which is really more about a licensing fee. Just using the word Bluetooth on your package or in your product requires an $8,000 fee.
I think someone in the academy had recommended one way around that is to just not use the trademark Bluetooth anywhere in your product, just say it’s a wireless instead of Bluetooth, but yes, all of these certifications tend to be costly. That’s another reason why it’s best to stick with one country and then expand out from there because otherwise, you’re going to be overwhelmed in certification cost.
Dave: And use pre-certified modules.
John: Definitely. Use pre-certified modules, wireless modules, specifically, that will definitely help reduce your FCC certification cost, allow you to get a nonintentional radiator certification instead of intentional radiator, a certification which– I’m not going to go into detail here, it’s something I’d spoken quite a bit about.
You can just see any of the blogs on certifications that I talk about that. Then, another one that I really think is a good way to lower your certification cost is if your product is rechargeable, just say it has a micro USB port.
I’ll see a lot of people present their idea and they want to design their own custom wireless charger or not wireless charger, but just a normal wall wired charger, but that’s another product. Now, that plugs into an AC outlet. That’s going to require UL certification, which is probably $10,000, and once again, you’re developing two products, two enclosures.
Get rid of that custom charger, instead buy just an off-the-shelf, you can buy them really cheap from Chinese suppliers, just a USB wall charger that’s already certified and just bundle that with your product and that’s going to not only lower all your development cost but also your certification cost.
Dave: Similarly, you never want to fall into the trap of looking at the cost of a wireless module and it’s $5 and your engineer says, “Oh, my God, I could do that for $1.” If your production volume is only 1,000 units, sure you might save $4,000 on the I’m not buying that wireless module, but what was your additional design cost, and most importantly, what was your additional certification cost and how long did it take?
John: Absolutely. Yes.
Dave: That’s a classic. Buy the module for the first rev and when the product’s successful, redesign it as a custom wireless module.
John: Absolutely. No, I think that’s a really good point. I think that’s it on cost. Was there anything else you wanted to mention on cost?
Dave: No, I think we did a pretty good job there.
John: I think so. We kind of wanted to just hit on some of the highlights there. We could do several podcasts just on cost, which I’ve done on development cost. Obviously, there’s development cost, scaling cost, production cost. Cost is obviously a very big topic and one that’s important. I think we’ve hit on the more important points. Let’s jump to number six.
Dave: This last one. You have a first-time hardware entrepreneur who does the right thing and goes to an experienced engineer for an estimate of time and cost.
They estimate that it’ll take, let’s just pick a nice round number, one year to take this to production and that year is made up of this much time in software and this much time in hardware and prototyping and debug, et cetera. That’s great. Now we have a schedule pinned to the wall, one year, 12 months, except there were some assumptions in that estimate.
The assumptions were, may have been, that the software would be designed on the eval board in parallel with the hardware and maybe the debug would be done by the same engineer sitting next to the software engineer debugging together. If you eliminate those assumptions, if you remove the ability for the software and the hardware engineer to sit next to each other because they’re in different continents or the software is done sequentially with the hardware instead of in parallel or any number of other aspects, then all of a sudden, that one year schedule is looking like two years.
John: Absolutely. I think there’s also assumptions that are being made, but also, just in general, I think most people tend to be optimistic, at least that’s an issue I always had in the past. I’ve gotten better over the years of estimating how long projects are going to take, but I’m an optimistic person and it’s hard for that to not seep into how long I think something will take.
You have to keep that in mind that rarely does– The schedule they may give you, maybe and is most likely very optimistic and that’s assuming everything goes smoothly, but almost never does everything goes smoothly and that’s where it becomes a lot more problematic to estimate how long something is going to take.
I can estimate pretty closely how long it will take me to design a schematic, a printed circuit board, but once those boards come back and any problems that are found, then that becomes a lot more problematic to estimate because now you’re dealing completely with unknowns.
If you knew about these issues, then you would have obviously not designed them into the product. I find a lot of schedules fall behind on that debugging stage of the project.
Dave: It’s much worse, as I mentioned before when the staff that did the software design, the hardware design, the firmware design, whatever it is is no longer available.
John: Absolutely. Kind of like that same point, it’s so much smoother and easier to estimate when all the engineers are under one roof. That’s why I’ve really, over the years, gotten where I– For some entrepreneurs, freelancers, can be a good route, but they require so much management and making sure all the pieces come together and just estimating how long it’s all going to take when you get all these different people working together, that’s really hard to do.
This is one area where I feel like if you get an estimate from a firm, then that’s probably going to be a lot more realistic, mainly because they also have a lot more data on how long things take and they’re more formal about how they do things, but just having everyone under one roof makes it a lot more easier to manage and to forecast how long it’s going to take to work together.
If your hardware guys are in the US and your firmware programmers are in India and there’s 12 hours difference then really when you’re doing debug and they have to communicate back and forth, that’s just going to really pile on the time.
Dave: Projects are rarely on schedule.
John: Yes, my experience at TI which is my main experience working with a big company on complicated projects, I don’t think I ever saw a single project complete on schedule.
Keep in mind this is a company that knows what they’re doing, tons of super-intelligent engineers on a project, lots of experience but it’s really hard to forecast and it would always happen when that first revision would come back, everything is on schedule but then things start to fall behind once problems are found in the design.
You’re trying to debug and it’s really difficult to know how long it’s going to take. I still remember one engineer I worked with at TI and I remember I think he got one of his projects back and he’s a really smart guy but he was so excited just telling me that, “Oh I think,” because he’d started looking at it. He’d spent a few days looking at it and then he was like, “There’s no problems. I think we’re going to have our fast pass success, we’re going to be ahead of schedule.”
Literally, a week later just all hell broke loose. Just all these major problems were found in the design and they ended up having to scrap it and basically start over, that’s how bad it was.
Dave: Oh my God.
John: Just keep that in mind, everyone is always challenged to try to get these schedules right so that’s why you have to be really pessimistic a bit on your schedule. I think whatever anyone tells you how long it’s going to take, even if they have a lot of experience then it’s probably still going to take longer than that and you may want to double that to be on the safe side.
Dave: You were just talking about a project at TI, if that project launched then it was a funded project, right? Whereas very often entrepreneurs are self-funding their designs which means that there might be a little bit of a delay at some point in the funding or while the entrepreneurs search for a lower course provider which could in turn also lengthen the schedule, right?
John: Absolutely, that’s a good point definitely. That can definitely lengthen the schedule as well.
Dave: The original engineer who did that one year estimate may have been assuming that it was fully funded and that the hardware and software could begin together sequential, I’m sorry in parallel. Where the entrepreneur has say limited a budget of 2,000 or 5,000 a month or whatever, in that case, it’s going to take a lot longer.
John: That’s a common way for entrepreneurs to be able to afford development as you have to stretch it out a little bit so you’re not doing everything at one time because then the cost can be a little bit overwhelming. Sometimes you just have to sacrifice time to mark it to be able to do that though, so it’s always a challenge estimating how long it’s going to take and you’re just going to have to get the best estimates that you can.
Then try to analyze all the assumptions that were made and from that forecast what you think is going to be realistic and probably even double that. I think that was the last thing we were going to talk about was projects taking longer than expected, is there anything else you wanted to mention about that?
Dave: You know John, one more that seems almost silly in retrospect because it’s not strictly technical is shipping time. You have an entrepreneur that’s done a great job and found low-cost providers in four different locations for the molding, and then the electronics, and then the assembly, and maybe the packaging.
Except there’s weeks of shipping in between each of those locations and all of a sudden, there was no shipping time originally included in the schedule. All of a sudden, there’s 12 or 20 weeks of total shipping time that’s got to be incorporated.
John: That’s a good one especially if you get all these different pieces spread out all around the world and you’re adding a lot of shipping time, that’s why it can be beneficial. That’s one advantage of China is they have a very large supplier eco-system that kind of keeps everything so most of the suppliers are going to be in the same country which can help to optimize that.
Dave: That’s all I had for today, John.
John: Okay. I think these have been six good points that we’ve talked about that entrepreneurs commonly underestimate. Let me quickly review them. Number one was the product is going to require multiple iterations to get the electronics right. Number two, it’s going to require updates over the life of the product.
Number three is it’s going to require prototyping on development kits or eval boards. Number four is multiple iterations are going to be required to get the enclosure right.
Number five was it’s going to cause more than you expect and number six it’s going to take longer than you expect.
Okay, Dave, this has been really, really great. I can’t thank you enough for coming on here for the third time and sharing some of the wisdom you’ve collected over the years. Can you maybe tell listeners how they can reach out to you and learn more about Dave, work with Dave, all that good stuff?
Dave: New address for me this time, come visit SalesDev.Global.
John: I’ll make sure I have that in the show notes, so that’s SalesDev, D-E-V, .Global?
John: Excellent. I’m sorry, so just SalesDev, just the dot between Dev and Global not between sales and Dev. Of course, Dave is also one of the most active experts in the academy so you can always reach out to him in the academy if you are a member.
He’s extremely helpful and is just in there pretty much every day answering questions and sharing his knowledge. See him @SalesDev.Global or you can connect up with him in the new academy. Dave, thanks so much for coming on. It’s always great chatting with you.
Dave: Pleasure to be here, John.
John: Okay, have a good day.
Other content you may like:
- FAQ: Which Should I Develop First: Electronics, Enclosure, or Software?
- Focus on the Big Picture for Your Product – Worry About the Little Details Later
- FAQ: Can an Arduino or Raspberry Pi Be Used in a Commercial Product?
- Overview of the Various Types of Hardware Prototypes
- The Minimum Viable Product (MVP) for Hardware Startups