Fluid Coupling

When exactly did enterprises become late adopters of technology? We know that they were some of the first buyers of computers. IBM sold tabulating and later computing machines to businesses starting in the 1910s. During the 1980s it was businesses which bought PCs in significant numbers to augment, and later replace, their centralized computing resources. Networking was in use in government and in business long before consumers saw any value in it.

In my talks I often point out that if you wanted to create a near-monopoly in computing in the 1990s all you needed was to convince 500 people to adopt your technology: the IT managers of the Fortune 500. If the largest companies used your product then they would impose the standard on all their suppliers and distributors and pretty soon there would be no alternatives.

So what happened during the last decade or so?

Today IT departments are known as the Information Denial department.  I recall that when the DVD first became an option on desktop or laptop computers, IT departments were first to decline the option (presumably because it would be used for entertainment rather than work.) When instant messaging first became available, it was IT departments who blocked the ports. When mobile devices with cameras became available signs went up that no cameras would be allowed on company premises. When USB sticks became available, USB ports started getting glued shut. When iOS became available, no devices running it were allowed on the network. Then came Facebook, Instagram and dozens of social media.

This pattern of not only a refusal to adopt but an outright ban on new technologies by enterprises made them fall off the radar of technology developers. Quite simply everyone outside the supply chain into enterprises stopped developing new markets around them. From venture funds to developers, enterprises fell out of the business plans.

The enterprise stood as a place of “legacy” and “security” which prevented mobile or other forms of computing. Paradoxes emerged wherein an administrative assistant had more computing power in his pocket than the CEO had in her data center; where the same assistant would know what was happening faster than any of the bosses. Homes had better connectivity than offices and productivity at small firms increased faster than at big firms. Incidentally, even the slowest enterprises were faster then the government. The bigger the firm, the slower and stupider it seemed. Were large firms employing dumb managers or did being a manager in a large firm make you dumb?

One resolution to this paradox might be that mobility and the movement of processing onto consumer devices increased the cadence of product development to such a degree that the purchase cycle and dollar amounts involved ran out of the range which companies could absorb.

A simple way to explain it is this: A company takes longer to decide to purchase a device than the device’s shelf life. In other words, by the time all the salespeople and committees and standards setting and golf playing and dining and site visits would be completed, the object whose purchase was being discussed would be discontinued.

A more onerous issue is that companies have procedures for accepting technologies (capital expenditures) which require high degrees of interaction and decision making. In order to step though these procedures, the vendors need to have sales people who need to invest lots of their time and therefore need to be compensated with large commissions. If those commissions are a percent of sale then the total sales price needs to be large enough “to make it worth while to all parties”. As a result, paradoxically, an enterprise technology must be sufficiently slow and expensive to be adopted.

Mobility was disruptive to enterprise because the new computing paradigm was both too fast and too cheap to be implementable.

This implies that the problem with enterprises is not the stupidity of its buyers. They are no less smart than the average person–in fact, they are as smart with their personal choices for computing as anybody.  The problem is that enterprises have a capital use and allocation model which is obsolete. This capital decision process assumes that capital goods are expensive, needing depreciation, and therefore should be regulated, governed and carefully chosen. The processes built for capital goods are extended to ephemera like devices, software and networking.

It does not help that these new capital goods are used to manage what became the most important asset of the company: information. We thus have a perfect storm of increasingly inappropriate allocation of resources to resolving firms’ increasingly important processes. The result is loss of productivity, increasingly bizarre regulation and prohibition of the most desirable tools.

Which brings us to the latest announcement of collaboration between the new disruptor of computing Apple and the vendors supplying Enterprises like IBM and Cisco.

Apple was the loser in the standardization of computing during the 1990s but is the winner in the mobilization of computing during the 2010s. The company positioned itself in both cases on consumer computing but it never gave up on enterprises.

The approach of Apple seems to be to enable the larger suppliers of technology to enterprises to bundle iOS as part of the acceptable set of services and products. In essence, Apple is complying with the requirement to be slow and expensive in order to be adopted. It can maintain the cadence of product development while attaching itself to the purchase cycle of the enterprise.

In a way it’s like an automatic transmission in a car. Operating through gears, the engine can rev at a different rate than the wheels turn. Occasionally, shifting happens but the fluid coupling keeps both the engine and the wheels from absorbing any damaging shocks.

  • Bill Esbenshade

    Brilliant post Horace, and I think very accurate — had not thought of the CapEx/speed angle before, but it makes good sense. Business people/users want top-notch mobile tools just like consumer buyers/users (and can communicate this priority to their IT departments), but corporate processes/systems make it impossible to keep up with the latest tech.

    I think your explanation is also counter to a lot of prevailing wisdom that “good enough” for an IT department buyer is lower than “good enough” for a consumer buyer/end user. Your explanation makes more sense to me. Business users want the same shiny toys that consumer users want.

    • v_sile

      “but corporate processes/systems make it impossible to keep up with the latest tech”

      Maybe, just maybe, there are some reasons. The enterprises are used to keep an electronic device (server,desktop, laptop, notebook, printer) for at least 3-5 years. Also they are used to get software update for their systems for about ten years or so.

      Changing expensive devices at every 16 months doesn’t look so appealing for enterprises. Maybe Apple or Samsung are very happy with this kind of life-cycle model, but the enterprises are not. Dealing with operating systems that have no idea what backward compatibility means doesn’t look so appealing, too. You have to change again and again your enterprise software applications. And that is expensive. Dealing with operating systems where security is just a fancy staff is a little bit problematic, too.
      In other works, speaking just for mobile market: consumer products are very good for consumers, but not for enterprise. At least not yet. The major requirements are quite a bit different. It is that simple.

      • Joe90

        The success RIM enjoyed destroys your main points, its that simple.

  • mrnowise

    How does one account for CapEX on information vs. CapEX on hardware/software? Maybe accounting rules shall be changed? or disrupted?

  • Tony Wittry

    Great insight… but incomplete conclusion IMO

    Good business leaders are dissolving and reconstituting their enterprises so fast that industries won’t even recognize themselves within the old style ‘enterprise sales cycle’…. And many of us ‘enterprise salespeople’ have been selling process and culture for a while now… :)

  • Jed Harris

    Another way to say some of this is that institutional capital expenditures have high transaction costs (in money, calendar time, and decision-maker hours, all of which impose opportunity costs). So there’s a painful cost-benefit trade-off to investing in low cost, rapidly iterated tech.

    This however raises the question of why these institutional heavyweight processes get applied to these low cost, low risk investments. The processes are not just preserved by inertia but are strongly enforced and defended. The internal institutional imperatives seem seriously skew to the larger institutional costs and benefits. I guess this is a case of principal – agent conflict.

    The arguments for these heavyweight processes are abetted by anxiety about “chaos” which is almost always overblown. The fear evoked is none the less real and quite powerful. It seems strongly related to anxiety about theft of trade secrets, feeding bad publicity, and similar issues which are also usually vastly overblown.

  • David Leppik

    “Paradoxes emerged wherein an administrative assistant had more computing
    power in his pocket than the CEO had in her data center; where the same
    assistant would know what was happening faster than any of the bosses.”

    I don’t recall this happening in the 21st century. Mobile devices still haven’t caught up with desktops/laptops, which in turn are less beefy than data center computers. Something similar did happen in the 1980s, when PCs often had faster CPUs than mainframes.

    Which brings up the real question: why DIDN’T this disruption happen in the 1980s and 1990s. Indeed, Moore’s Law type disruption was fastest in the 1990s, when clock speed could be increased by using more energy. (Eventually they came too close to speed-of-light constraints.)

    In the 1980s, there were home PCs and office PCs. The office PC took off, the home PC didn’t. I would argue that the reversal of fortune has little to do with Horace’s specific points, and more to do with the fact that personal—rather than work—computing finally started to encroach on work computing.

    • just a theory

      Why didn’t it happen in the 80s/90s? Great question. Corporate IT buyers don’t want to get fired. Security holes get them fired. Un-productivity gets them fired. Enabling the organization to be faster is intangible and doesn’t get them recognized.

      As networking proliferated and things moved mobile and to the cloud, security threats proliferated. My guess is that going from no computer to computer in the 80s/90s was so productive, the way to not get fired was to make this happen. And in the 2000’s, CYA has shifted to making sure these super productive things called computers don’t break or get threatened by outside parties.

      • Luis Alejandro Masanti

        I also think that most of the IT decisions taken during the 80 and after were done on ‘fear.’ It was typical to say: “No one get fired if you buy IBM.” Microsoft took later the post.

  • fstein

    Great article, thanks.
    I was consulting with Cisco during their Cius (android phone) launch. It fit your scenario perfectly, with down rev OS and chips; And priced way above comparable Android devices. Cisco wisely and quickly exited.
    Still we should give companies like IBM and Cisco credit for they do. They solve the tough problem of compliance – with their own legacy products and with regulations, security etc. Compliance feeds into the overhead, often to excess. It is the quintessential necessary evil.
    HP is the big loser to IBM and Cisco. Maybe… after HP breaks up into enterprise and consumer, the enterprise company can make a deal with Apple. Good luck.

  • Walt French

    “Today IT departments are known as the Information Denial department.”

    As they have for some time now:
    http://dilbert [dot] com/strip/2010-05-08 shows the mechanism in action.

    Mordach the Preventer of Information Technology hasn’t appeared much lately, perhaps only because he’s so fully integrated into the company workings that it’s not worth calling out.

  • Panos

    There is also the role of bureaucracy per se (regulations, standards, complexity etc) in IT acquisition. In governments things are even worse (some large companies resemble governments institutions).

    “The government has to follow a code called the Federal Acquisition Regulation, which is more than 1,800 pages of legalese that all but ensure that the companies that win government contracts, like the ones put out to build, are those that can navigate the regulations best, but not necessarily do the best job. ”

    “Federal IT procurement.
    Procurement is the term for the government’s Byzantine procedure for buying things. Under the current system, the federal government buys technology the way it buys battleships: with broad “Indefinite Delivery / Indefinite Quantity” (ID / IQ) contracts that pre-screen vendors to handle an agency’s projects for up to 10 years.
    Because the procurement process is such a headache, agencies often lock in contractors for longer periods. This speeds things up, but it also gives preference to Beltway insiders and excludes smaller companies. As a result, new programming frameworks and development methods take a long time to reach the government. A company that has already bagged a 10-year contract has little incentive to innovate.”

  • Klajd

    I believe short term strategies of some managers, CEOs play a significant role as well.

  • hannahjs

    Fine engineering analogy! Was the clutchless transmission in any way “disruptive” of industry or society, when it became widely adopted back in the 1950s?

  • ptmmac

    As a small business owner, I am seeing more and more tech adoption by slow moving businesses and government agencies. Not fast enough to make it easy, but fast enough to offer several different ways of doing things. The capital costs of integrating new mobile apps to old databases and structures are slowing adoption, but not stopping it. The oldest application I run is my register system software in my restaurant. The security issues of handling credit cards, and the lack of immediate incentive to upgrade the systems that were designed 20 years ago had until recently made things a big head ache or needlessly expensive. The iPad register system that I tried, BreadCrumb, had a lower cost upfront, but much higher costs hidden by lack of careful integration of peripherals, many missing pro level features, and cloud side network dependance. The credit card readers were much slower, more prone to break and reaquired network access to handle transactions. Other teething pains included menu programming that was too easily done with out requiring that reports balance or having proper trouble shooting tools.

    My register system, Aloha, an RDS product, was upgraded to Windows 7 and required 30% more capital input to get all terminals and servers upgraded. The payoff was a drop in maintenance costs by 15% and access to new cloud services from NCR who recently purchased RDS. If Apple had made the new rumored iPad Pro available more quickly, then perhaps the services available such as computing power that servives internet outages by running it’s own internal server would have been sufficient to make the change over work. NCR is raising our costs by charging more for cloud services than Breadcrumb. That is not as big a deal when our telephone costs are dropping enough to make that part of our current IT budget.

    The real factor that old tech has over new tech is it already has been fully debugged. Something that works smoothly has a value that can’t be matched by buggy improvements. Design choices that made sense for one simple type of business software, become limitations when applied to larger more complex organizations. The real choice is not slow and expensive versus fast and nimble. It is between slow, expensive and marginally functional, and fast, new, partially functional systems. Hopefully, the new iPad Pro will fix some of the consumer system limitations of the current iPad. If it does it will disrupt PC’s at a much faster rate than the iPad has. Perhaps the old truck versus car idea that Steve Jobs offered is a bit disengenuous. What we have with iPad is a truck with no bed, rather than a car. It can go anywhere a truck can, but it can’t carry proportional loads.

  • Jack Schofield

    There’s actually nothing very new about this phenomenon. In the 1980s, I had a conversation with Michael Dell, who said that some of the PC orders he got from large corporations were for models that had already been discontinued. He said something like “We just send them the latest model and a refund…”

  • obarthelemy

    I’m really not sure this is unjustified, or new, or actually the case.

    – there are a lot of valid reasons to hold back on integrating the latest consumer techs into an entreprise infrastructure, from ROI to security to long-term lock-in toa single vendor in the case of Apple… What’s your criteria to say they should move ahead more quickly ?
    – if you look at it in a biased enough way, there have always been techs that haven’t been adopted quickly by corps: inkjet vs laser, wifi vs wired, latest Windows vs older Windows, even PC vs mini/mainframe… what’s your criteria to say now is worse than before ? Not enough Apple purchases ?
    – there are plenty of techs that are enterprise-first, in storage, networking, virtualization, GPGPU… what’s your criteria to say enterprises are lagging in tech adoption ?

    • Kizedek

      Once again, it sounds like you are the biased one. It’s pretty clear Horace is talking about mobile/”post-pc” technologies and purchases vs “tradtional/incumbent/more slowly developing” technologies — he’s not talking about MS vs Apple purchases, per se.

      For example, it’s pretty hard to rapidly adopt a new point-of-sales system using tablets and a new payment system (whether Apple or Android Pay), if your whole organization is built around buying and amortizing cash registers (not to mention the physical placement, installation and maintenance of shop counters and conveyor belts), plus Windows to operate the cash registers (which has its own lock-in and licensing system that is far more expensive and onerous than Apple’s, BTW).

  • Ray

    Interesting insight on how software and personal client devices don’t fit traditional CapEx processes.

    I feel though there are a couple of missing important points:
    – Enterprise buyers are mostly rational actors, well educated in their areas, that need to justify their purchase decisions to executives, CFOs, CEOs… based on RoI (maximize jobs-to-be-done per dollar)
    – Consumer buyers are not as well educated in the area of purchase (most car buyers are not experts on cars, most smartphone buyers are not experts on IT) and choose based on many other factors besides RoI (design, retail experience, advertisement, etc.)

    Apple, as a consumer company, is a master at pulling the levers that make consumers tick, but those same levers don’t work with more rational and knowledgeable enterprise buyers. Dell competed on price, delivery times, customization, etc, which are things enterprise buyers care mostly about. If Apple is really serious on penetrating the enterprise PC market, they need to focus on those. Smartphones are a bit different as they are cheaper than PCs and RoI considerations are less important as providing the workforce with choice.

    • Bruce Hobbs

      ROI is outdated. Although it can be a factor in evaluating a new technology, relying on it as the sole means of evaluation will kill a company. Today, there are so many intangible costs and costs of avoidance. What is the cost of losing a AAA bond rating? Do you know? I don’t. Did you know that the cost of a security breach is usually less than the cost to prevent it? Where’s the ROI in sufficient security to prevent a breach? There isn’t any!

      • Ray

        I think what you’re arguing is that RoI is difficult to calculate, not that it’s outdated. Essentially calculating RoI means trying to forecast the future: the investment is now, the return is in the future.
        RoI has always been difficult to calculate in any field (marketing spending, R&D, military spending, etc.) but that doesn’t mean it’s outdated. All organizations try to estimate the RoI, including intangible costs, which is better than just spending money without considering the estimated return.

  • TomCF

    I’m amused that every article you post has comments that show exactly how people and organizations behave that perfectly prove your points.

    • Horace Dediu

      That is the value of a comment system.

      • TomCF

        Not every site does so well.

    • katherine anderson

      Horace is not always clear himself on his points, but that’s part of the point; his commenters, and the friction among them (like any creative process), help to develop the point … ideas sometimes lead in unexpected directions and surprising places.

  • Jeff G

    Way better than… Fluiding Couple