innovation failures

Boeing's Big Boo-Boo: Top 6 Innovation Takeaways (Part 3 of 3)

In this 3-part series, we dissect the failure of Boeing's inflight satellite-based internet service, Connexion by Boeing. In parts one and two, we discussed the flawed business model and massive marketing mistakes that inevitably led to nearly $1 billion in losses before shutting down the project. In this 3rd and final article, we ask what could have been done differently, and what lessons should be learned.

Recap: A 6-Year Long String of Failures

After looking at the cost structure, revenue models, poor assumptions about market size and demand, pricing, scale of operations and marketing strategy, the inescapable conclusion is that almost everything that could have been done wrong to bring Connexion by Boeing inflight internet service to market, was. Like so much other big company innovation, it was clear that Boeing had no idea how to go about launching and driving market adoption of this service, or how to design a business model that had a chance of working.

Part 3: What could they have done differently?  (Hint: Next time pretend you're a startup.)

Although not in any way a disruptive innovation, or even one with potential to disrupt, the Connexion by Boeing service could have learned a lot from how disruptive innovations are successfully brought to market and how startups launch products.

Here are some of the things they could have done differently and lessons that can be learned from Connexion's failure.

1) Business Model Design

The cost structure that this service was saddled with, from satellite bandwidth licensing to having a business unit with 560 people to incredibly expensive technology needed on each plane, boxed Boeing in. To get the service available in a large enough number of planes and at the right price point to make it economically viable, Boeing ought to have used a strategy more akin to the early dotcom successes like Amazon and Yahoo and Google.

If your product depends on a mass market to be affordable, then you have to seed the market to bootstrap it. That means pricing low -- even giving it away for the first year (like Skype is doing) to get people on the grid and addicted.

Ultimately, the pricing strategy needs to recognize what the average or downmarket consumer considers good value, not be a skimming strategy dependent on only those willing to pay the maximum.

Lesson 1: There are many possible variables in designing a business model, and what works in one scenario doesn't always work in another. Don't assume you got it right the first time, and don't scale your business before you know that it's viable.

    2) Fast Failures

    Boeing assumed that the service they wanted to deliver at the price they wanted to sell it for was the service that the potential customers wanted and were willing to pay for. There was no small scale testing of this before scaling the business into a 560 person business unit, locked into very expensive satellite bandwidth contracts, targeting availability anywhere in the world.

    Fast failure is a necessary part of startup methodology. The objective is to test the constraints, minimum viable product features and alternative business models, and assumptions about the target market. You expect to fail multiple times but learn from each failure until you arrive at the right configuration or determine that there is no product/market fit or viable business model.

    Big companies tend to see failures as something to penalize rather than something to learn from, which means when they fail, they fail big, just as Boeing did.

    Startups without the luxury of Fortune 500 cash backing them are forced to do things in a more lean fashion, and to test all their assumptions before committing them to the business model. This little bit of common sense might have been enough to work out the kinks without blowing a billion dollars, maybe even hitting profitability within months of operation.

    Lesson 2: Failing is a good thing in a startup if you do it fast, learn from it and pivot. In fact, it's a necessary component of identifying the right business model design and optimizing for success. It also costs a lot less than assuming you're right and failing slow.

    3) Cost Structure

    Lean startups need to keep their cost structure pared to the bone until they demonstrate that they have a business model that can scale because VCs wouldn't have it any other way. As noted above, Boeing built an outrageously costly platform making it virtually impossible for the service to be profitable unless every unrealistically optimistic projection came true, and boxing themselves into a price model that customers wouldn't buy into.

    Disruptive innovations tend to have the lowest cost structure, often by an order of magnitude. Boeing choose to implement almost the highest possible cost structure. This left the service vulnerable to lower cost technology advances (e.g. onboard conventional wifi from ground transmission stations and/or plane-to-plane daisy-chaining) that could be expected to compete in the future, undercutting and undermining Boeing's business model.

    Lesson 3: Always try to achieve the lowest possible cost that delivers a product/service that is "good enough" to satisfy your target market, and be conscious of technology advances that could obsolete your innovation using cheaper components or lower manufacturing and assembly costs.

    4) Network Effect Potential?

    One of the factors that strongly favors disruptive innovation is the existence of a potential network effect. In this case, there was a weak potential network effect that could have been exploited, which favored convenience and usability and faster time-to-critical-mass.

    Had Boeing promoted Connexion directly to consumers with a message that they could now get affordable (assuming the price was affordable, which in their model, it wasn't) internet access in the sky, using a single account with a single sign on and promoting the carriers that offered it, and gained a large subscriber base quickly, the other airlines would have to get on board or be left out of the game.

    Achieving this potential network effect / sole supplier status depended on having the right business model to begin with, and getting out of the starting blocks fast once product/market fit was established.

    Lesson 4: Always ask whether there is a possible network effect, and if there is, what you need to do to enable it. Don't get greedy in the early days, because networks can easily stall before reaching critical mass if you overprice or enable alternatives to establish viability.

    5) Customer Development

    The key idea here is that ideas about customer needs formed in a laboratory and without talking to customers are hypotheses, not an accurate description of real needs. The only way to uncover real needs and gain an understanding of who the market is and what its members are willing to pay for with a service of this nature is to go out in the field with a prototype/working model and observe and talk to them. Customer Development is the methodology most commonly used by successful startups to establish what the real needs are and get to Product/Market Fit.

    It appears that Boeing failed to understand that the airlines weren't really their customers, just gatekeepers. Unfortunately, for aircraft sales, the airlines are their customers, which complicates this enormously. Boeing probably shouldn't have been in this business, or should have found a way to work around this fundamental conflict.

    For this service to be successful, Boeing needed direct access to the end customer -- the passengers using the service. They needed to have a few market-test planes where the service was fully available, but could be run experimentally with different ways of constructing pricing. They needed to sit beside customers who were deciding whether or not to use the service and to see directly everything about its operation and the user experience, from how the customer pays to price barrier resistance to difficulties signing on to how it was communicated by flight crew to passengers.

    These and many more attributes needed to be known so that an affordable, easy-to-use, properly communicated and desired service could be introduced. Because this wasn't done, none of the obvious objections about usability, availability of power outlets, space to work, price point, etc. were encountered or thought of, creating numerous barriers to adoption.

    Lesson 5: Understand who your real customer is before you launch a service saddled with $1 billion in cost structure and have no way to reach them.

    6) Establish Product/Market Fit

    Establishing product/market fit is the process of identifying the compelling need that your target customer must have satisfied by the Minimum Viable Product, as well as all the assumptions that this is based on and the risk factors.

    If you haven't spoken to real customers, it's impossible to know whether your product/service innovation meets any compelling needs that they are willing to pay for. It's also difficult to uncover the risk factors that will inhibit adoption, or to know which assumptions are valid, and which are invalid.

    Boeing theorized that passengers were willing to pay for inflight internet access, but never validated that assumption with more than a flawed market survey, nor properly established if sufficient demand could be created at the pre-determined price point to have any chance at economic viability.

    They also didn't observe firsthand customers choosing not to use the service because they wanted to sleep, or to watch a movie, or because they lacked sufficient space to get value, or didn't have access to power or sufficient battery life -- they never learned any of the risk factors so that they could act to address/negate them, or pivot their strategy and work around the risks.

    Lesson 6: With any innovation, you must know how customers view the product, what the risk factors to viability and adoption are, what alternatives exist (to using your product -- they don't have to be direct competitors), what segment(s) you appeal to and what will motivate them to a purchase. You have to get out in the field with customers (not imagined customers) and see it yourself.

    Other Issues

    • Traditional market research instruments and focus groups are useless for designing non-sustaining innovations. They cannot uncover real needs and market drivers. The most you can expect from research with end-users is that they tell you their pain, not a solution or its value. After they've tried it in a real scenario, they can give lots of feedback about what's wrong with it. The conclusions of Boeing's research indicate that the survey used was an invalid research instrument that asked irrelevant self-serving questions - the sort used to fill press release content, not to determine customer needs.
    • Accessibility and Convenience. Every plane needed to be outfitted ASAP for this service to take off. Amazingly, it was not until this year that the first new plane was built and delivered with Connexion installed. Boeing should have made the decision to include it as a standard piece of equipment, and charge more if the airline said no. This way, there is no retrofit cost barrier to laggard carriers deciding to offer the service -- they could even conduct trials without having to plan years ahead and make big commitments.  Furthermore, the cost to retrofit existing planes should have been made much more attractive, even going so far as to bundle it with major maintenance service intervals at cost of materials, or cost of labor, whichever was lower.
    • Take responsibility for your own marketing. If you need to work through the carrier to reach the customer, do their marketing and business planning for them. A Connexion-in-a-box kit, because if they have to think, you've lost the game.
    • Pricing. What the market has clearly said regarding any communication/network service such as this (think landlines, cell phones, earth-based broadband, Tivo, even text messaging) is that it wants a reasonable (read: relatively low) subscription price which is capped, and which an average user perceives as good value. For example, if it costs me $30 for 2 hours of access, I'll wait until I'm on the ground and connect for free at the airport lounge (not free really, but I already have a subscription that works there). Very little email is that urgent, and I certainly won't pay that for gaming or to watch TV over the net.
    • What about the lack of carrier vision? Indeed there is plenty to pin on the carriers too. They don't get marketing either, at least not in a user-oriented authentic context. For any of the US majors, this would have been a cinch to create brand loyalty, much like frequent flyer miles did when American was the only one doing it. Think of this: a premium flyer package, as an add-on to my Crown Room or Admiral's Club membership.  How about a $15-20/month upgrade to my membership that not only gets me Club access, but gives me unlimited internet access on the plane, higher priority waitlisting for upgrades, a free drink on domestic flights and two freebies on international (if I'm stuck in cattle class), plus phone calls, movies, the whole shebang. It's hard to imagine anyone not buying this, with an associated increase in demand for Club memberships and an extremely sticky preference for a ticket on that carrier. Ironically, I would probably pay $30-40 more for a ticket on a carrier where I had those privileges, and that level of increased ticket price is well within most business traveler's discretion at time of purchase.

    In the end, it's about being sensitive to customer needs and perceptions, and offering a product and packaging that removes barriers to adoption, rather than erecting them.

    And carriers, please, power access in every seat should have been done 15 years ago.

    Boeing's Big Boo-Boo: A Very Non-Disruptive Innovation (Part2 of 3)

    In this 3-part series, we dissect the failure of Boeing's inflight satellite-based internet service, Connexion by Boeing. The first article looked at the laboratory-designed business model that was untested before scaling to business unit size, but based on made-up assumptions and faulty analysis.  In Part 2 of this series, we look at the marketing mistakes and myopia that contributed to the colossal failure.

    Recap: An Unworkable Business Model

    In Part 1, we discussed how Boeing was so sure that this innovation couldn't fail that they invested more than a billion dollars over 6 years with only $25 million in offsetting revenue before folding their tents. We can't say for certain that anything was learned because at the end, Boeing maintained that the tragic events of 9/11 were their undoing, even though clearly this was designed as a "Spruce Goose" from the outset.

    The business model was truly unwieldy. Depending on super high-cost satellite service, Boeing needed more than 10x the number of planes that had been outfitted by the time the program was shut down to have the necessary equipment installed, and had they achieved that level of penetration, more than 40% of all passengers on every single flight (assuming the flights were 100% full) needed to pay full price for service just to break even. And, it cost $500,000 per plane to install, and the pricing assumption was that passengers would pay more than the cost of their monthly broadband service at home per hour in the air.

    Moreover it required a presence in the US market where internet usage is still the highest and the majority of commercial flights begin and end. But even on the carriers where Connexion was deemed a success, usage rates were in the single digits, and Boeing never changed its stance or its business model to encourage greater use.

    It brings new meaning to the phrase "what were they thinking".

    Part 2: Counting the Marketing Mistakes

    "Extensive" market research" by Boeing came up with a definitive statement that "38% of frequent travelers are willing to pay at least $25 per flight for full, high-speed access to the Internet and their corporate network". Clearly, Boeing believed that it they built it, the users would come, and had boxed themselves into a cost structure that embedded this assumption.

    Which leads to the first marketing failure: a dependence on MBA-styled market research. Asking dopey questions has never been shown to result in better odds of being successful at innovation than a coin toss would yield (at much lower cost, but then it doesn't seem as scientific), and in this case, not only was the research instrument flawed, but the wrong question was asked.

    Consider this: if I had an urgent deadline that absolutely couldn't wait until I landed, I might be willing to pay $25, so I'd probably answer yes just to influence the outcome (so that the service became available), however, the real question should have been "at this price level, how often would you use the service", just to mitigate the built-in bias.

    And, whatever answer you got, you'd test it a dozen different ways for validity. At this ridiculous price level and with this billing model, I might use Connexion once or twice a year, but more likely never. Certainly on flights from the US to Europe, most of the time in the sky is overnight, so who is waiting for my urgent communications anyway?

    Ignoring all that, and the fact that I represent the prime target market, the really head-scratching fact is that Connexion's business model required between 30 and 40 percent of all passengers on all flights to pay for this service every time they got on a plane, and nearly 10x as many planes carrying the service as had been outfitted.

    What's Wrong With This Picture?

    What really stands out is the lack of a "gut check". No comparison seems to have been done to the pricing models that have worked for similar services, the rate of adoption that they experienced, or when the growth rates exploded and what triggered them.

    Moreover, the obvious comparison to another overly high-priced satellite communication service that was highly touted but also failed miserably, namely Iridium, was clearly missed. Admittedly, a "gut check" is less quantitative, but it can be immeasurably more accurate.

    So, they got the pricing wrong, but that's not the only factor that was going to limit uptake. Marketing apparently never considered any of these barriers to adoption in their business model.

      1. Loss of sleep time. Because most of the travel time on US to Europe flights is overnight, that means the service is competing with sleep time on at least half the potential flights. If I have a meeting first thing on arrival, best case is that I lose several hours of sleep time because of the number of time zones crossed, so unless I'm desperate, I'm highly unlikely to even open my PC. Even if I do work on my PC, the constraint on my available time to use this service is so severe that the cost seems even more ridiculous.
      2. Other activities competing for my time.  Movies, food, bathroom breaks, getting up to stretch, drinking, reading, listening to music, watching a DVD, talking to my travel partner, doing PC work that doesn't require a connection - there are numerous things to do that I don't have to pay anything extra for.  As my colleague Mike Urlocker is fond of saying,"Don't trust what people say; trust what they do".
      3. Scoble notes that most flights don't have access to power. Oops.  Unless I'm in business class, I'm pretty much assured of not having enough battery power to get full value for my connection fee, and even in business class, not all planes have power connections, nor is it easy to know ahead of time if you will have power or not. In the grand scheme of things, as battery life gets longer and longer (mine will go 5 hours at near full power) this is a minor irritation, but it is definitely a nuisance factor that will prevent potential users from plunking down their cash.
      4. Desire to be off the grid. Scoble also notes that many of the techno-geek frequent flyers at Microsoft (if they don't want it, who does?) get some welcome relief from being out of touch for a few hours on their flight. A feeling we can all identify with, I'm sure. And, at the outrageously high price tag, anyone who might otherwise use the service has a good excuse in the high price for not bothering.
      5. Lack of space. Ironically, economy class on long-haul international flights is much more cramped than on domestic flights. I've been on flights where I couldn't get upgraded, and the first thing the passenger in front of me did was recline his seat causing multiple fractures in my kneecaps. As anyone who's been on an older L1011 or a 747 knows, the space between rows is too small to fit in even if the passenger in front of you doesn't recline. Assuming the seat in front of you is in upright position, it is nearly impossible to work on anything other than a micro-laptop, and as soon as the person in front wants to sleep, you not only can't work, your circulation gets cut off too. This is perhaps the strongest incentive to not risk paying the fee and not being able to use the service.
      6. Poor promotion by airlines. In the initial 3-month market test by Lufthansa, the first and most successful airline to offer the service, they claimed an average of 50-80 simultaneous users per flight, 95% of whom were happy with the service. How does this jive with the actual results after launch. How often do you try something once just to see whether it's worth it? Most of us do. But after the first time, value, accessibility and awareness become much more important. Also not disclosed is how many of those users simply logged on to Lufthansa's FlyNet portal, which was a free service. 180 flights per day is a tiny fraction of the actual flights in the air - Boeing noted in their initial announcement of Connexion that there are 41,500 flights worldwide each day. If you were going to be on one of the 0.4% of flights that had the service installed, did you even know it was available and plan your time accordingly? Unlike all the press release hoopla, there was little product differentiation going on in the market, and even less promotion to make people aware when they were going to be on an internet-enabled flight. Even at the gate, there was often no posted notice that you were boarding a Connexion flight. Boeing depended on the marketing efforts of the airlines, and the airlines attitude seemed to be "if we have it, they'll figure it out and use it." At least with Lufthansa, you knew what you were getting because they commited to install Connexion on all long haul flights, and implemented relatively quickly, but their average users per plane (if they were all paying customers) was still only 1/2 what Boeing needed across the system to succeed.
      7. Lack of uptake by US carriers. When the very first announcement of planned service was made in June 2001, Boeing said that American, United and Delta Airlines had signed letters of intent to adopt this service and equip 1,500 planes with Internet connectivity. Of course, on 9/11 that commitment evaporated, and to date, no US-based airlines have signed on. This is a bit of red herring, since if this had gone ahead and all 1,500 planes were outfitted by now, there would still be only 10 times the number of planes that currently have the service, and we've already pointed out that they needed 400x the number of users to make this fly. However, it would certainly have increased awareness, and possibly led to price drops or more innovative business models that would have given this a chance to reach critical mass.
      8. Lack of direct communication with end-users. Connexion considered their customers to be the airlines, although they were clearly dependent on very large scale adoption by the real customers - the end-users. In an utterly amazing statement of 'not getting it' Boeing said "Each of the airlines brings an unprecedented level of knowledge about the in-flight connectivity needs of passengers." as it announced the pending partnerships with US carriers.  Right.  Just like they understand my need for sufficient room to sit, and my need to pay a fair price for services, and my need to be treated like a customer rather than a filled seat. Boeing, as a near monopoly provider to US carriers also didn't get that if you want a mass market to develop, you can't dictate the terms.

        And, I didn't even have to think hard to come up with this list of show stoppers. I'll bet there are others, but since I don't need to think that hard to make the argument that product marketing didn't even do a half-assed job, I'll leave it at that.

          Bottom line: the market research was garbage, the assumptions the business model was built on stank, and almost everything about how this was packaged, priced and marketed was wrong. Even if the US carriers hadn't dropped out of the picture, it is highly doubtful that this service had a fighting chance.

          In Part 3

          In parts one and two, we look at the absurd business model and financial assumptions, as well as the "big dumb company"-styled product marketing and lack of gut check that led to the stillborn Connexion service market failure. In the third and final part, we ask "what could they have done differently", what lessons could be learned, and whether any other approach could have succeeded.

          Boeing's Big Boo-Boo: A Study in Self-Disruption (Part 1 of 3)

          In this 3-part series, we dissect the failure of Boeing's inflight satellite-based internet service, Connexion by Boeing. In part one, we examine the faulty analysis that resulted in a business model that was impossible to execute. The type of innovation represented by the Connexion service, and how the business model came to be is typical of large-company thinking, and why disruptive innovation so rarely comes from incumbent industry players.

          Part 1: A business model that couldn't take flight

          On July 17, 2006, Boeing announced the discontinuation of its in-flight "hi-speed" internet service.  A quick analysis of the business model shows that this dodo was doomed to extinction before it took its first flight.

          Why it had no chance - by the numbers

          Connexion offered a satellite-based internet connection which enabled internet connectivity even on transcontinental flights. Started with plenty of fanfare 6 years ago, Boeing's Connexion service won the "World's Leading High Speed Inflight Internet Service Provider" award from the World Travel Awards organization in London for 3 years running.

          Said Connexion by Boeing president, Laurette Koellner, "Winning this award for the third consecutive year is a welcome validation of our success."  That was in November 2005, just a scant few months ago.

          Exactly what success was being validated though?

          Service Usage

          According to data released by Boeing, "more than 20,000 passengers have used the Connexion by Boeing service during its first year of availability."  20,000 certainly sounds like a big number, but as an absolute number, it is free of context and that context is important. Anyone who knows how to make statistics lie would agree that the best way to distort the meaning of figures is to offer them without context, or juxtaposed in the wrong context.

          In truth, 20,000 was the only number that Boeing could have published that made things look positive. In a separate announcement, Boeing offered other data which tells a more complete story. 

          A corporate backgrounder published a few months later in early 2006 shows that Connexion was available on 180 flights daily. That is 65,700 flights per year, give or take a few. 

          Assume that a few more planes were outfitted between the time of these two announcements, and perhaps the run rate was 60,000 flights per year at the time Boeing said 20,000 passengers had used the service. That is one person on every 3 flights using the service. Not such a stunning success anymore.

          So, lets be generous since we don't know the rate at which service was added during the period when Boeing had 20,000 users. Maybe it was actually 1 person every two flights. Hmmmm. Still not much of a business model.

          Revenue Streams

          Now let's look at revenue, and assume that everyone who used the service paid the top daily rate during that first year of $30. 20,000 paying customers at $30 each is a grand total of $600,000 revenue.

          We know, however, that prices were reduced during this period, and many of these passengers likely opted for the lesser priced packages, so lets give a nice round number and say they had $500,000 in air-time revenue from end consumers. This for a technology that cost over $500,000 per plane to install. I'm starting to smell some really rotten fish. (140 aircraft had been outfitted at the time this backgrounder was written.)

          Of course, there were other revenue streams from government, ship traffic, corporate jets, branding fees from airlines but it's clear that this business model and the cost structure was based on hefty uptake from business travelers. The WSJ says Connexion had $25M in revenue, but also notes that over $1B was invested over 6 years.

          Profit Potential

          What about potential for profitability? We know that there were very high fixed costs associated with this service including expensive satellite bandwidth purchased from several providers, roaming agreements with land-based providers to ensure smooth logon and hand-off of communications, and payroll for 560 employees, the majority of whom required expensive technical knowledge to do their jobs.

          Boeing didn't break out Connexion separately in its operating statements, but lumped it into a category called Other which was mostly Connexion. In the most recent quarter, Other showed a loss of $90M for the quarter (improving from a loss of $110M in the quarter a year earlier).  Let's be really generous (and for the sake of nice round numbers) and say that only $50M of that loss is directly attributable to Connexion.

          Again, for the sake of round numbers, let's say that the losses due to Connexion were approximately the same every quarter, or $200M annually. To generate $200M in revenue and break even, Connexion needed 400x the usage they were getting at last year's higher prices. (Since these are quick calculations, we aren't considering that higher usage would also mean higher expenses).

          At that rate, instead of 1 person every 3 flights, what was actually needed was an average of about 130 paying customers on each and every flight.

          Where Was the Gut Check?

          That stunning calculation leads us to the first great "Aha". What marketing/business genius put together a plan that could only succeed if a very large percentage of the target customers on every flight had so great a need that they would be willing to pay almost as much for a single flight as the cost of a month's broadband access on the ground?

          Even if there are some Fortune 500 companies that would be willing to allow that as an expense, I certainly wouldn't want to be justifying more than 1 per month of those to my boss or the corporate controller. I might be able to get away with it, but it's a nuisance factor I don't need.

          Another indicator that these round numbers aren't so far off: Boeing anticipates $0.15 per share benefit to next year's results coming from this decision, which at the current number of shares outstanding is over $113M in increased earnings. Since most of Connexion's staff will be moved into other jobs at Boeing, and therefore the majority of payroll expense will continue, that seems about right.


          • why was there no business model experimentation to determine whether the market as envisioned would or could ever materialize?
          • how were the usage, revenue and profitability projections determined?
          • how did Boeing manage to scale this service to require massive fixed costs and 560 employees to support the service before determining product/market fit and the real size of the likely market?
          • were any of their assumptions realistic, and if not, how did they miss such a gaping hole in their business plan?
          • can a big company realistically behave like a startup to test business models in "fast-fail mode" before scaling?

          In Part 2

          In Part 1, our quick analysis of the accounting clearly shows the magnitude of the miscalculation implicit in the Connexion by Boeing service. It looks as though it had no chance to reach profitability ever. So, how did this come to be? In Part 2, we examine the marketing mistakes that contributed to this colossal failure.