Chapter 30:
Death of a Product
In the spirit of full disclosure, I probably should have put this chapter before the previous one. The events outlined here occurred in the early to middle part of 2013, whereas our “worst customer ever” came a bit later. But when I originally outlined this book, this chapter didn’t exist at all—mainly just because I was dumb and forgot about it.
However, this doesn’t make this chapter any less important. It covers something that any business will have to face at one point or another: the death of a product.
Of course, most product deaths aren’t usually deaths, per se—they’re more a phoenix-like event, where a new and even shinier product rises from the ashes of the previous one. At least hopefully.
But…that leads us into the first question. When a product gets long in the tooth, do you update it, or give it the full Kevorkian treatment?
Good question. And, with that, let’s talk about product life cycles, and product life cycle management. Yeah, good boring corporate stuff. But I’ll try not to make this too tiresome.
Product Life Cycles, AKA the Game of Update, Assassinate, or Cannibalize
Okay, let’s start with the basics:
- No product is fresh or competitive forever, especially a technology product; the competition, and the market, can and will change—sometimes in new and unpredictable ways.
- Because of this, you have to think in terms of product life cycles—or, in regular English, how long a product will be a good, solid competitor in its market.
- You should determine (at least) a guess as to how long your product life cycle is, so you can be working on updates or replacements before the end of its life.
And, the bonus stuff that most companies ignore:
- Killing your babies is perfectly OK, if updating won’t make them a good product for new market realities—you have to be ready and willing to do this.
- It’s better for you to cannibalize your own product lines, rather than waiting for someone else to do it.
The above is why you typically see an iPhone every year. It’s why most other flagship phones are on the same life cycle—the technology, software, and market have changed enough in a year that a new, fresh product is required to stay competitive.
It’s also why you see new laptops and such on timeframes dictated largely by the release dates of new chipsets from major manufacturers like Intel—the introduction of the new chipset changes the game enough so that new products need to be introduced.
Cars? They have longer life cycles, typically 2 years between minor refreshes, 4 years between “making it look new on the same platform” and 8 years between moderate to major platform changes.
Same goes for a lot of less techy stuff—appliances, etc. Their product life cycles can be much longer than a year.
Audio? Hmm, now that’s a conundrum.
On the mass market side, the major manufacturers of “bulk” gear like receivers and such have been chasing a 1-year product lifecycle for a very long time—but the new products that come out frequently aren’t anything more than re-badged and slightly de-contented versions of their predecessors.
Aside: My theory is that the ongoing decontenting allows them to maintain arbitrary price points (determined by copying their competition). After all, it’s much easier to follow somebody elses’ rules than make your own. The latter might require intelligent marketing to show how your products are, well,
actually different from, and better than, the competition.
Another aside: And that’s assuming they
are actually different and better—which may not be a good assumption in the mass market.
In high-end, product lifecycles are all over the place. Some companies make essentially the same products for nearly a decade. Some make changes every couple of years.
What’s right?
- Some use the old metric of “when sales slow down, it’s probably time to update.” But this is an astoundingly bad metric. When sales slow down, it’s probably too late. When sales slow down, you’re under pressure to come out with something, fast. And when you’re under pressure, you may miss a critical feature—or not do your best work.
- Some use a fixed schedule: “We have decided our product life cycle is 3 years, so we will begin working on significant updates 24 months after launch.” This is better, but what happens when the market undergoes rapid changes and your sales fall off a cliff at 18 months in? Do you simply sit and wait the remaining 18 months to launch a new product?
When we started Schiit in 2010, if you’d asked me what I thought our product life cycle would be, I probably would have shook my head and said, “I don’t know. Two years, three years? Let’s see how it goes.”
And, to this day, I can’t really put a number on it. We’ve decided to set our product life cycle on a more flexible metric than falling sales or dates on a calendar. If I had to put it in words, it would be something like this:
Our products are updated or discontinued when significant positive changes can be made, or need to be made, at a time not disruptive to customers.
Note the specificity: updated
or discontinued (it’s okay not to keep a zombie product alive), significant
positive changes (not just a small tweak, and not de-contenting), at a time
not disruptive to customers (updating a product 3 months after release, for example.)
In the case of Asgard, early in 2013, those significant positive changes
needed to be made.
Why? One word: Magni.
The Death and Rebirth of Asgard
Before Magni, we never really thought about updating Asgard. It was a great amp, a strong seller, and sales continued to pick up. By the “wait until it slows down” metric, Asgard was doing fine.
But as soon as we heard Magni, we all looked at each other, and said, almost in unison: “What about Asgard?”
Magni was just too close in performance to Asgard. In fact, it was more powerful than Asgard. It was quieter. It ran cooler. About the only thing Asgard had going for it, objectively, was a much better volume pot—better tracking, better taper—you can’t beat large pots for that, no way, no how. Aesthetically, it was a much more elegant-looking piece, but elegance only gets you so much. Sonically, we believed it was still ahead of Magni—but not by enough.
So it was time to look at the end of Asgard.
But did we kill it, or did we update it? That was the first question.
Deciding to update it was really easy:
- Asgard was still selling well after Magni—in fact, sales continued to increase until it was discontinues—so there was clearly still a demand for a step-up amplifier.
- Magni had taught us a lot about surface-mount parts. Surface-mount parts, applied to Asgard, would actually reduce its manufacturing cost due to robotic assembly. This would open up putting the money into other areas.
- We regularly got requests for two features Asgard didn’t have: gain switching and preamp outputs. Adding these made a lot of sense.
- Assembling Asgard was a real pain, with four separate MOSFETs, plus thermal pads, screws, lockwashers, and electrically insulating bushings for each—I’d been thinking about a new way to do it with one simple bracket, two screws, and a single custom thermal pad. We were now of a size where we could do the large custom orders that made the switch viable.
- Asgard’s gain stage was pretty basic—it didn’t even have a current-sourced front end, and it had coupling capacitors at the output. Mjolnir had taught us a lot about how good a high-voltage, cascoded, current-sourced front end could be—and going to surface mount meant we could afford to do it—together with a trick DC servo that connected to a sonically innocuous (as in, non-amplified) terminal to eliminate the coupling caps.
- And while we were at it, we put the rest of the money saved by surface-mount to change the transformer and power supply from unipolar to bipolar on the output, plus added an 80V rail for the HV front-end like Mjolnir.
That’s a big list of updates, guys. Different board, different topology, different power supply, different chassis, different transformer, different feature set—the only thing that our proposed Asgard 2 had in common with the first generation was the same chassis styling, the same pot, and the same basic connections.
So how did development go on this radically updated Asgard?
Almost comically boring.
I built a perfboard proto one evening and verified that the basic concept worked—including delivering measured distortion performance that was almost 10X better than the original Asgard. Total time: a couple of hours.
From there, the first boards we got fired right up and worked. They even fit the revised chassis just fine. All the pain of your typical product development—all the tiny little nits and problems—were conspicuously absent. I changed a couple of resistor values from the initial calculations, but that’s about it. Total time: maybe 4-5 hours.
From there, we took the first in-chassis prototype inside to the listening area where I had Mjolnir and Gungnir set up. I plugged it in, put on some headphones—most likely HD800s, because they’re great at revealing what’s wrong with an amp—and sat back.
Crap, I’m still running the Mjolnir, I thought, as soon as I heard it.
But I looked over…and the headphones were plugged into Asgard 2.
Asgard 2—sounding like a Mjolnir?
No way. Not believing it, I switched the gain to Low, and the volume decreased, as expected.
Holy moly, I
was listening to the Asgard 2.
Still not really believing it, I swapped back to Mjolnir—and, yeah, Mjolnir was a step up, but it wasn’t leaps and bounds like the original Asgard. Of course, Asgard 2 didn’t have the raw power for, like, HE6s, but it was
very good—much, much better than the original.
Next, IEMs. Dead silent on low gain. I grinned. This was gonna be a winner.
I decided to call in Rina and demo it for her.
“I want to hear the Asgard 2, not the Mjolnir,” she said.
“You are.”
She went through the same rigamarole as I had—switching gain, unplugging headphones, comparing to Mjolnir—before believing it.
“This ain’t no Asgard,” she finally pronounced.
Mike? Pretty much the same reaction. We had a winner.
And, you know, sometimes things just work out. And sometimes, things work even better than you expected. In product development, this is known as a “gimme.” Also known as, “oh crap, watch your back.”
I should have watched my back.
The Asgard 2 Launch Debacle
The runup to Asgard 2 launching was filled with the same little delays that happen with every new product—waiting for parts, waiting on chassis, etc—so, in that respect, there was no clue that we had any nasty surprises waiting in compensation.
But when we launched the product, word quickly came back—some of the Asgard 2s hummed like a refrigerator. Mechanically. As in, you could hear the transformer humming with headphones on. Closed headphones.
But that made no sense—none at all. The prototype hadn’t hummed. And we hadn’t heard any hum in production.
“But we wouldn’t necessarily hear it,” Alex said, as a big train went by outside, shaking the paper-thin stucco of the Schiithole.
“Crap,” I said, realizing for the first time just how loud it was in our building. It wasn’t just the trains—it was the constant traffic noise from cars passing on the 4-lane road outside.
Late at night, we confirmed it. Many of the transformers did hum. And the prototype didn’t.
Why?
This is known as a “production surprise.” As in, “Surprise…although the transformer meets specs, we decided to make them a little differently…and that difference transformed your product from a headphone amp into a combo headphone amp/massager.”
The transformer manufacturer was apologetic, and promised new samples posthaste. But that didn’t fix the humming Asgard 2s in the field—now the entire first run. We’d been so deep in backorder, we’d sold out in just a few days.
So what did we do? The only thing we could do: accept the returns on the Asgard 2s that hummed, or swap them out as we got the new transformers in.
Aside: gimmes are dangerous, guys…be suspicious, be very suspicious, of something that is too easy.
And that’s why we started a new policy after Asgard 2: multiple prototypes, multiple listeners…and multiple locations, some of which we knew were quiet.
Yeah. There you go. But that’s also why the launch of Valhalla 2 and Lyr 2 were, well, relatively uneventful. Almost boring.
And, in terms of “production surprises,” that’s exactly where you want to be.