Interesting article - but I can't help but feel it's written with a big dose of 20:20 hindsight. Obviously Iridium didn't succeed and if you were doing it again and knew what you know now, you would do things very differently (as the article explains). However, back in 1990 when Iridium started, I think a lot of their decision were not necessary ignorant.
In 1990, Iridium probably did have a good solution to a problem in 1990. The article mentions that the waterfall model was a bad development model to follow - yes agree - but whilst 'agile' development concepts existed back to the 60s, it really only started becoming more mainstream in mid 1990s (the term 'agile programming' wasn't even used until 2001). So the waterfall model was probably a reasonable model to take back then.
"No Business Plan Survives First Contact With A Customer". Again in the age of release early, release often, this seems to make sense. But when your product is global satellite coverage, how do you break down your product into something that could be released early/often? I guess you could possibly launch one satellite and see if people would subscribe ... but like the Net, a one-node network would be of very limited value and reach.
I agree that the frozen business plan was the critical reason in their failure. Even a regular survey of potential customers would have shown that the price difference was too great for rapidly diminishing benefits.
But I don't agree that their initial solution was necessarily bad or that the waterfall model was inherently bad based on 1990s knowledge.
>I guess you could possibly launch one satellite and see if people would subscribe
For that, though, you'd need to put the satellite in geosync (nobody is going to subscribe to a phone service that's usable only every 100 minutes), which costs more to launch and requires the phones to have stronger transmitters. Thuraya uses geosync, but they started out as a regional carrier, so they could cover their region with one satellite.
Iridium went with polar LEO--I suspect that was partly because their goal was global coverage. Without a polar orbit, you can't get coverage at the poles. Insert cynical laughter about the profitability of such coverage here.
I don't think he's criticizing them for not following agile methods, considering they didn't exist then. He is criticizing them for not changing their plans based on the developments going on around them.
But mostly, he's just using Iridium as an example of what not to do, regardless of whether they could have known better or not.
I think he is exactly criticising them for not following agile methods - it's the basis of his criticism for following the waterfall method. His criticizm for not changing their plans is pt 3 (which I agree with)
And on your last point, using Iridium as an 'example' is like criticizing Middle Age 'surgeons' for not using antiseptic to ensure the success of their operations. You'd be right, but it's hardly a valuable observation nor provide much insight into how to avoid an equivalent situation today (where you are lacking knowledge/tools/technology).
You'd be surprised what information was available vs what was actually being done. Lots of stuff we think are modern discoveries have been known for 2000 years.. yet not widely practiced for various reasons.
I guess it boils down to what is lately being called "evidence-based medicine". Trial and error would have taken them really far, if they'd practice it properly. Which is something even now we're not doing to the fullest extent.
Iridium never identified a customer need, they just ploughed ahead and built the biggest, shiniest system and hoped that it would pay off. Global coverage just isn't that useful to all but a tiny number of users.
The obvious customers were seafarers, but they had been using Inmarsat for years. Their network achieved near-global coverage using four geosynchronous satellites. You needed a big antenna and a powerful transmitter, but that was no problem on an ocean-going vessel. The same went for fixed installations in remote areas on land. Inmarsat was at the time a UN-backed nonprofit established to improve safety of life at sea and was under no particular commercial pressures.
GSM took off in a huge way in the early nineties, so by the time Iridium launched most of the developed world was blanketed in GSM coverage. Multi-band handsets and roaming agreements meant that most travelling business customers already had a handset that worked just about everywhere at relatively low cost.
There was still a gap in the market for portable communications in less developed and more sparsely populated areas, particularly Africa and the Middle East. The later network Thuraya solved that problem cheaply and elegantly by putting a geostationary satellite over Ethiopia and covering most of the market for handheld satellite phones after just one launch. They now cover 70% of the world's land mass with just two satellites and are far more profitable than the revived Iridium.
Iridium could have pivoted. They had a lot of sunk R&D costs, but the initial 66 satellites were built and launched in an enormous binge between 1997 and 1998. They had a gap of at least six years between conceiving the idea and building the first piece of hardware. I believe that they failed to do so because they were preoccupied with the idea of providing truly global handheld telephony, rather than the more mundane task of making a product that people want. Iridium brilliantly solved the technical problem of providing telephone coverage to parts of the world with very few people, but failed to realise the basic flaw in that business model.
Iridium turned out to be a wonderful gift to humanity, but it was a bloody awful business decision.
I'll bet that if any Iridium engineers are reading this they are shaking their heads in disbelief.
Once the hardware decisions were made they were locked in. The lead times in aerospace are nothing like software or consumer electronics. You can't exactly tweak your hardware when it's orbiting the earth at 500 miles.
By the time they launched they must have known of the shortcomings for several years but it was too late to do anything about it. The best they could do is launch and hope that a big enough niche would emerge to fund further development. Sadly, that didn't happen.
It was a bold, daring plan that achieved spectacular results. They were very unlucky it didn't work out commercially - even if the reasons are obvious in hindsight. I don't see how they could have done any better.
Who knows, maybe one day somebody will figure out a novel way to capitalise on it.
In the 80s when only yuppies had cell phones and coverage was limited to a few square miles of financial districts could you have predicted 100% GSM coverage in Africa?
There still isn't 100% GSM coverage in Asia - the most densely populated continent, let alone Africa and S. America. There are many many places that will never see GSM coverage. So I wouldn't write satellite communications off just yet.
Could you have predicted it in the 80s? Maybe, if you didn't have all this money to build exactly what they built.
Compared to the cost of installing conventional telecoms Iridium's $6Bn looked very cheap.
Their big mistake was assuming that mobile phone usage would be a limited special market alongside landlines. They didn't consider that much of the 3rd world would simply skip over the cost of installing landline infrastructure and cell phones would be the cheap alternative so leading to a much larger demand and cheaper phones and calls.
Agreed. One thing missing from the article is that the Iridium satellites only offer a paltry 2.4Kbps for data communications. In 1997 and 98 the business world was still largely on dial-up. Internet access on the road was patchy and incredibly expensive. If they'd pivoted to global data coverage at even 28-56K, I think they would have had a far more significant take up.
I like this article because I see examples of this mistake all the time.
There is a startup (forget the name) who is seeking to bring smartphone like features to the much larger feature phone segment of the market (~80% of the market). It's a big idea and the potential market is huge... on paper.
The problem? Technology is ever changing. Today what are $500-1000 devices (and thus the top 20% of the market) will in 2 years cost half that and be 50% of the market. So they're betting on a business that by the nature of pretty much anything to do with silicon, is a dying business. Basically all phones will be smartphones in the next 5 years.
In the digital camera market, Olympus and some others designed the FourThirds sensor format for digital DSLRs some years ago. They did this at the time when Canon and Nikon were making larger sensors and putting them in $1200-2000 bodies. Their idea? Reduce the cost of the sensor and have something that is compelling both in terms of price but in body and (in particular) lens size.
The problems? APS-C sized sensors became basically a commodity item. So FourThirds has only a size advantage and not really a cost advantage. Canon and Nikon (in particular) have much larger markets and the economies of scale that really matter when it comes to lenses.
So the lesson of the article is don't bet your business in what is a transitory problem.
While I don't disagree with the basic idea you're presenting, about people building nice sounding solutions that end up DOA, your Four-Thirds example is a bit distracting, since it isn't true.
Olympus and Panasonic have made a lot of money selling four-thirds format cameras, for advantages quite unrelated to sensor cost. These two companies carved an entirely new market segment (very small high-quality-sensor exchangeable lens cameras). Previously there was no exchangeable lens digital camera that was both small and optically (sensor and lenses) high quality. What's more, in the traditional segments Olympus was a bit player with sales very disapponting compared to the Canon juggernaut. So this story is much more about an old player deciding to play an entirely different game rather than just keep getting the stuffing beat out of them by just going along competing directly.
They've done well enough that other players (Sony, Samsung) are jumping in and trying to pitch their own mirrorless formats to try to win a piece of this new market segment. There is speculation Nikon and Canon will try as well.
You're confusing the Four-Thirds format with the Micro-Four-Thirds (to be fair, they have the same sensor area).
The Four Thirds format - i.e. traditional SLRs with smaller sensors was a failure until they pivoted into Micro Four Thirds (EVIL cameras, "electronic viewfinder interchangeable lens"). The cost advantages of four thirds sensors never materialized, and the smaller sensor size is only a marginal advantage when it comes to integrating into smaller devices.
This is evidenced by the fact that a mere year after the launch of the first Micro Four Thirds cameras, Sony has already hit the market with the same concept, but they're doing it with a larger APS-C sensor (and also no bigger than the E-PX and GF-X series of Micro Four Thirds cameras).
Olympus and Panasonic rightly deserve the credit for inventing the entire concept of EVIL and bringing it to market, but IMHO it has little to do with the Four Thirds format. That whole thing was a bad idea to begin with. APS-C sensors are widely available and commoditized, so there are no cost advantages, and there is not necessarily a significant technical advantage either.
I was going to try to just continue this via email, but I don't see your contact details. I don't have the industry background to know whether APS-C sensors are indeed "commoditized" but that sounds a bit spurious. Canon, Fuji, and Nikon make the best digital camera sensors in the world today, and I have not seen a single camera produced by a major player that uses one of the other player's sensors. They'd be insane to sell that advantage as a commodity part. In that environment, it makes a lot of sense for Panasonic and Olympus to push the development of their own supply of sensors.
As for whether there is nothing worthwhile in their smaller sensors, well, time will tell. Today, there is no MFT-competitor (Sony, Samsung) lens with Aperture below F2.8. That's pretty half-hearted competition.
The main advantage is that optical aberations increase quickly with sensor size (upto diameter^4). That's why your cell phone is able to take pictures with such a crappy lens.
What Olympus did was exactly the opposite of the article - they gave themselves the ability to make lenses that were much smaller and cheaper, or much more ambitious in performance for the same size/cost - by gambling that semiconductor progress would make smaller sensors just as efficient as large ones.
But how exactly do you know for certain that a problem is transitory without the benefit of hindsight?
Should Apple have never started with iphone development because Nokia/Ericson were rapidly innovating in the phone space? Iridium back in 1990 thought they had the equivalent of an Apple iphone in their solution.
I don't actually think the article contributed much in the way of insight. Without trying to be rude, your last sentence not to bet a business on a transitory problem is fairly self evident - the real value is if you can outline how to avoid stepping into such a situation without the value of hindsight (which the article fails to do).
I think the Iridium situation in 1990 was a limitation of silicon chips. Moore's Law pretty much guarantees that a problem caused by limitations in silicon will be transitory problems.
In the case of the iPhone, there was a lack of innovation. The silicon of the day was up to the task. It's just that nobody had a product like the iPhone. The closest you'd find would probably be a Blackberry.
it went into a 8-year Waterfall engineering development process. Waterfall development is a sequential way to develop a product (requirements, design, implementation, verification – ship.) Waterfall makes lots of sense in a market with the customer problem is known and all customer needs and product features can be specified up front. It is death in a rapidly changing business
Paraphrasing a well known smuggler, launching satellites into orbit ain't like dustin' crops, boy.
I think the point Steve is making is that as the company went along, they should have noticed their prior assumptions about the cellular market were not panning out. They cold have started seeking out alternative revenue streams besides traditional dialup voice service. I'm not sure what those options would be, maybe messaging or some other data? Two-way Internet via satellite? I'm not a telecom person, so I'm ill-suited to come up with workable options.
The problem is that Irridium was sort of an all or nothing design. You needed the whole fleet of birds to have global coverage. Geo-sync birds might have reduced the count but they cost more to launch and have a very noticeable lag.
I'll make a different assertion, at the time they dreamed it up, the business model had no chance of working. There were no pivots or re-adjustments. There weren't a lot of great data options at the time. Most importantly, if you launched half the fleet and then decided to re-tool for data support or something else, you'd have partial and inconsistent coverage during that retooling and then you're data coverage would be partial and inconsistent.
The gamble was voice was good enough and global coverage was a big enough seller that they'd re-coup the investment and become one of the largest telcos in the world and then by the time the birds died they'd launch next gen ones. It's a dynamic equation, there is tremendous demand for voice, unfortunately there isn't tremendous demand where there is no service.
The only "agile" choice they could have made was to stop. The real choices for the business were to wait (but they had money lined up, how many businesses raise the money and then call it off?) or to go forward. Personally, I just think it couldn't have worked, wrong idea at the wrong time. No agile vs. waterfall debate can change that or make a bad idea a good one.
But what could they have done differently? They could perhaps iterate the design of each satellite on the assembly line. But things like buying rockets, getting them delivered to the launch site, getting a launch window, etc, it's hard to see how you can do that in sprints. This is about moving physical stuff around.
The point is "they should have done something differently" not "what they could have done". Halfway through their development process, the problem they started out to solve no longer existed.
Launching satellites etc are expensive and time consuming. Agreed. But even when Iridium launched it's satellites were not fully operational. They would have probably had enough warning to iterate on a different product
We don't know that they had a choice. If you've signed a contract, with penalties, for 15 launches, you might know you're doomed but have to press on anyway. The Eurofighter is another example of this - when the UK govt tried to cancel its Tranche 3 order, it found the penalties would have been more than the cost of the aircraft...
I'm not really sure what the lesson is here. He says "Customer Development, Business Model Design and Agile Development could have changed the outcome" but how? I mean, they could have shut down halfway through and saved a couple billion, but it doesn't seem like any amount of adaptability could have made a global satellite project cost-feasible in the face of cell phone ubiquity.
The only lesson I can think of here is don't finance a project that's going to cost a lot and take a decade to implement, because there's a good chance that technology will make it obsolete by the time you launch.
"The only lesson I can think of here is don't finance a project that's going to cost a lot and take a decade to implement, because there's a good chance that technology will make it obsolete by the time you launch."
I think the lesson is to be continually re-evaluating the project to see if your assumptions still make sense given the current state of the market. This would be especially true for multi-year projects. It's basically the fallacy of sunk costs.
And wouldn't that have been a huge improvement over the actual outcome? Shutting down early and saving 3 Billion dollars is 125X what they ended up selling for in bankruptcy.
I agree with you in principle. That would have been a better outcome. But as they say, hindsight is always 20:20. When you're in the thick of things, the last thing you want to do is quit. Sometimes you only want to quit when you have to, so that at the end of the day, you know you gave it your all.
It's an improvement, but not much of one. If the best you can figure out in hindsight is a way to only lose $2 or $3 billion dollars, you're not trying hard enough.
If you realize your market is niche, would you build different satellites than what they did? Change the economics of it somehow (cheaper phones, lower cost per minute, longer life on the sattelites etc) ? Make a satellite/ground station hybrid that can work through most walls? Shrink the phone size?
The big one would be: Only launch enough enough satellites to cover the area where your main customers are, rather than trying to cover the whole planet.
That's a little tricky - the earth being round and all.
You can't put a satelite in orbit around New York (in spite of it being the center of the universe - orbital mechanics doesn't work like that)
Your only real option is how close to the equator you orbit them - you can get away with about 3 sats if you only want to cover the region very near the equator, unfortunately only poor people live there.
As you improve coverage further north you have to add more satelites in higher inclination orbits. Iridium already limits this - the coverage in the far north is very poor, which is ironic since it's one of the few places where you really need sat phones!
One change they could have made is add data to the satellites. Data was greatly desired by the users, but the system was not designed for it. The best they could do was a slow analog modem.
"The only lesson I can think of here is don't finance a project that's going to cost a lot and take a decade to implement, because there's a good chance that technology will make it obsolete by the time you launch."
I generally hear about Iridium as a second-generation success. The fact that the company that currently owns it is looking at putting up a new network (at a cost of perhaps $3 billion) shows that the idea wasn't a foolish one.
But like some great ideas it takes a trailblazer to burn themselves up in the trailblazing process.
Another example in this context is the Panama canal. Or various railways companies. Or possibly the Segway (still waiting for the second generation owner on that one).
Regardless, I'm sure happy that mistake happened. When the main satellite goes down on the ship I work on, the Iridium phone is the always-working standby you can count on.
The current Iridium looks like it has focused its marketing to the segments that actually make sense: maritime, explorers, and military.
But the infrastructure is up there and, sure, a bunch of people lost a lot of money, but it's THERE now. I could think of some products they could introduce that might be promising.
Here's an idea: a small piece of plugin hardware for iPhones, Droids, etc. This would be something you take with you (for example, on road trips or vacations to far-flung places). You would buy the unit for $149 and then prepay, say $30 for 10 minutes of service. The idea is that it's "there when you need it, wherever you are." It could even go in your first aid kit... it's there for emergencies.
Commit 5 billion dollar to a 'frozen' business plan, that will take years to execute, while the world changes under your feet. You know there's no room for iteration, so you don't pay heed to whats going on until you've burned all the cash.
Bliss in ignorance that was by design because of the nature of the business plan.
Things to avoid doing: spending crap-tons of investment money long before your product is on the market; gold-plating the business plan; relying on the classic "expect everything to go as planned waterfall / big-bang integration" model.
Iridium could have been much more successful (read: profitable to the initial investors) if they had used different techniques at each step such as iterative development with strong market / customer feedback. If they had begun by gauging market reaction through using pre-order commitments (paid and unpaid) instead of relying heavily on deep pocket investors they would have been able to scale their initial offering to the market a bit better. If they had planned to launch a "minimal viable product" initially that could be scaled up over time as the business grew they would have avoided the massive up-front costs that sank them (the biggest cost was launches, they could have pre-built the entire fleet of satellites and kept most on the ground for a pittance in comparison).
Instead they put an entire fleet of satellites into orbit at tremendous cost and launched a service which on day 1 gave essentially 24/7 phone coverage over the entire Earth. An impressive achievement to be sure, but at too great a cost.
Some companies can grow like this, ramp up to enormous proportions quickly and find enough traction to keep running and keep growing. But for every success there are many more failures, companies that devoured their seed corn too fast.
I wonder what Iridium could have done better, if they used more agile methods and followed the market. Sure, they could have discovered it wouldn't work half way though, given the remaining money back to investors and closed down. But what company does that?
I cannot really see a good "minimum viable product" to test the market for Iridium either. Nobody would buy a phone that only works for an hour each day, when the first satellite in orbit is overhead.
This reminds me (on a smaller scale) of Realtime Worlds spending 5 years developing the APB game and then having a cold shower this year when the game was not so successful as they thought, following a close down. (http://en.wikipedia.org/wiki/Realtime_Worlds)
> The idea of iteration or pivots was unthinkable. This business plan was a static document. It was great for fundraising, looked great in business schools
He (unintentionally?) implies that business schools are similarly clueless, which is funny since he teaches at Stanford's.
It would be interesting if they managed to cut the ties to Motorola completely after the spinoff? One can imagine the BigCo mindset prevailing even after leaving the mothership.
Optical would have all the problems the current RF (Iridium) solution has plus more...
* Clouds are a killer.
* Laser is highly focused - downlink would be tricky to support more than a couple of ground stations. Uplink would be extremely difficult because the uplink would have to track a (relatively) fast moving LEO satellite. Neither would lend themselves to a handheld ground station ("phone").
* Non-laser solutions likely don't have the bandwidth or the power to be effective.
Indeed, aiming the collimated beam would be a problem, but not an insurmountable one. Given the bandwidth benefits and the relatively small size of the transmitter (compared to microwave), many small (cheap) satellites could serve the same purpose as an Iridium constellation.
It wouldn't work indoors, but neither do satellite microwave links. The reason cell tower signals reach indoors is signal strength. It's prohibitively expensive to place equivalent transmitters into orbit.
UV wave generation requires significantly more energy consumption than microwaves. Plus, UV radiation is highly absorbed in the atmosphere, leading to low signal to noise ratio.
Various super high bandwidth dod constellations (TSAT which was recently cancelled due to costs) included optical inter satellite links in space. Basically, a few gbps of connectivity dorm each satellite to the ground via rf, and then laser links in space at 10-40 gbps linking the satellites.
Not really as feasible for the ground link for a variety of reasons, but the space part actually makes sense.
In 1990, Iridium probably did have a good solution to a problem in 1990. The article mentions that the waterfall model was a bad development model to follow - yes agree - but whilst 'agile' development concepts existed back to the 60s, it really only started becoming more mainstream in mid 1990s (the term 'agile programming' wasn't even used until 2001). So the waterfall model was probably a reasonable model to take back then.
"No Business Plan Survives First Contact With A Customer". Again in the age of release early, release often, this seems to make sense. But when your product is global satellite coverage, how do you break down your product into something that could be released early/often? I guess you could possibly launch one satellite and see if people would subscribe ... but like the Net, a one-node network would be of very limited value and reach.
I agree that the frozen business plan was the critical reason in their failure. Even a regular survey of potential customers would have shown that the price difference was too great for rapidly diminishing benefits.
But I don't agree that their initial solution was necessarily bad or that the waterfall model was inherently bad based on 1990s knowledge.