In the past ten to fifteen years, many barriers between traditional
industries have broken down. We're in the early stages of another
big, blurry brawl, but to set some context, here are a few examples
and data points:
-Entertainment and computing now overlap in many significant ways.
According to Neilsen, Americans between 8 and 34 spend more time
gaming than watching television. Globally, computer gaming has become
about a $30 billion industry, compared to worldwide box office
receipts of about $26 million in 2006, which was an all-time record.
-Telecommunications and media are battling into each other's
territory. Cable television and voice providers, as of 1990, were
separate and distinct. By 2000, cable providers led telecoms, in
North America anyway, in Internet access. Starting early in the
decade, the cable providers have gathered significant share
(approaching 30% in some markets) of the wireline voice market, but
DSL has gained back some share in Internet access. In the next five
years, look for telecoms to provide television content; Verizon has
announced Fios1, a "hyper local" channel in the Washington, D.C. area.
A recent study by Motorola found that 45% of Europeans, led by 59% of
the French, watch some TV over the Internet. Cellular telephony is
shaping up as the next media platform: Japanese phones routinely
include television tuners already, and growth is expected to be rapid
in many areas of Asia.
-Retail has been redefined along several dimensions. The U.K. grocery
chain Tesco punches far above its weight in petrol sales: with only
4.3% of the retail locations, it has captured over 12% of the market,
lagging only BP, which controls 16.5% of the market but has three
times as many locations. Here in the U.S., it's hard to believe that
Wal-Mart expanded from general merchandise into the grocery business
less than 10 years ago, but it controls at least 20% of a highly
fragmented market. eBay, which began as a secondary market, now also
includes many new, branded goods from established sellers.
Our focus today, however, is on a different "invasion" of an adjoining
market. Ten years ago, when investors were looking for "the next
Microsoft," they held certain assumptions about what a highly
successful software company looked like:
-Winners choose the right platform, picking well according to the
market size and share of the hardware on which the software, of
whatever sort, runs. IBM's OS2 operating system in the early 1990s
had some technical advantages over Windows, but IBM never established
the application layer which would make its OS competitive.
-Winners develop mechanisms for user lock-in and network effects.
Word processing programs stand as an obvious example, where switching
is hard and expensive, and it makes sense to be on the same product as
all of your co-workers.
-Winners manage upgrade cycles efficiently: that locked-in user base
will eventually have to buy the new, improved version, delivering a
major revenue infusion to the software seller and perhaps the wider
-Winners sell software to large customer bases one consumer or one
business at a time. This reality of the market implies effective
management of brand, retail channels, and enterprise sales forces.
-Winners care about software functionality and performance; data as it
is generated or managed by the application falls out of scope.
-Winners hire strong technical teams because functionality is
specified early in the new product development cycle and must be
hard-coded into the package.
-Winners think of the world in rows and columns. Whether in
calendaring, spreadsheets, databases, project management (swim lanes),
presentation graphics, or customer contact management, most programs
of the 1985-2000 period deeply embedded a grid metaphor and/or
Times have changed. Whether one looks at Google, a clear challenger
to Microsoft's dominance, or at the new crop of companies all seeking
to ride the "web 2.0" bandwagon, many of these assumptions about
software no longer hold true. For example, in a survey of 25 startups
to watch compiled by Business 2.0, fully 20 had revenue models at
least partly based on advertising. Greatness in software now requires a lot of the old world skills and positioning, plus a healthy dose of some new elements as well.
To set some context, look at some familiar companies listed by market
capitalization and price/earnings ratio as of 19 July:
Company Cap P/E
Microsoft 301B 23
Oracle 105B 25
SAP 67B 26
Disney 68B 16
Time Warner 78B 13
Google 171B 48
Yahoo 35B 51
Ten years ago, anyone looking for "the next Microsoft" probably would
not have looked to Viacom or Disney as models. And for good reason:
the role of "pushed" content is itself in transition. Yet the core of
the media model -- the packaging of audiences for sale to advertisers
-- is fueling growth at Google, presenting both technical and cultural
challenges at Yahoo, and the source of deep concern among Microsoft's
top leadership. The changing of the guard is further emphasized by
Microsoft's experience with the most recent exemplar of old-school
software, its Vista operating system. The product shipped three years
late, with a stripped-down feature set, and effectively cost several
senior executives their jobs: Brian Valentine, Jim Allchin, and, to a
degree, Bill Gates. It also has yet to sell in large numbers, in part
because enterprise buyers are waiting for the first updated release,
when many of the first-run glitches will be addressed.
What are the emerging dynamics for software dominance? Compared to
the standards for success circa 1997, a few factors have been inverted
while most still hold true, with a twist.
Rather than developing for Unix, Windows, Mac OS, Symbian, set-top
boxes, and a variety of other operating systems, Google and Amazon
have led the way toward development of services for the Internet as a
platform. Among other things, this stance greatly simplifies product
distribution: the differences between Google Maps and my 1998 version
of Rand McNally's Windows package are striking. Every time a new road
is built, or interstate exits renamed, or a pedestrian mall built,
millions of CDs become obsolete. Google (or NavTeq or whoever) makes
one change to the base map and every subsequent query will be addressed with accurate information. Getting the platform right still matters, but the definition of the term is changing from local to virtual, solitary to distributed, and product to environment.
This aspect still concerns financial analysts, particularly because
switching costs can be so low. If I change from Yahoo Finance to,
say, Fidelity's investor workbench, apart from my investment in the
old interface, there's very little to restrain me from leaving. Tim
O'Reilly, who helped formulate the very notion of Web 2.0, asserts
that users own their data in these sorts of scenarios, but the
exceptions to his assertion prove that Web 2.0 is hardly the last
word. My eBay reputational currency, iTunes preferences, and Hotmail
account are neither open nor portable -- by design.
There's no question that successful software still exploits network
effects. The more developers who code to a given platform --
Facebook, Salesforce, or Google maps -- the more that standard gains
authority: note that none of those aforementioned businesses counts
as a website. One of the platform pioneers powerfully illustrates the
point perfectly: Amazon just noted in its earnings conference call
that it has 265,000 developers signed up to use its web services.
There are also powerful network effects among users, whether at eBay,
MySpace, or BitTorrent: the more people who use the service, the more
valuable it becomes. Compare that one fact to consumer products,
banking, automobiles, or pharmaceuticals, and we are reminded how
significantly online dynamics depart from those of widget business or
even most of the service sector.
Upgrade (and therefore revenue) Cycles
No longer is the objective to leverage a large installed base onto a
new version of the product. Google makes money every hour of every
day, and apart from acquisitions, we don't expect spikes in its
revenues. Indeed, the escape from the cyclicality of product upgrade
cycles may not yet be fully appreciated as analysts assess the new
breed of software companies. The dependence of shrinkwrap software
companies on secondary revenue streams may become problematic: Larry
Ellison noted in an interview with FT last year that Oracle was
getting 90% margins on maintenance. Customers can't be, and aren't,
happy with those economics, so it is likely only a matter of time
until competition and/or customer resistance change the model. To what, nobody can say.
Selling Software as a Product, One at a Time
On July 19, Google reported quarterly revenues of $3.87 billion, a
year-over-year improvement of 58%. Did its sales force grown by 60%
in a year? I highly doubt it. Although the company offers a few
software products a customer can purchase, they amount to mere drops
in that $15 billion annual bucket: enterprise search hardware and
software, hosted applications, GIS tools. An important facet of the
(lowercase) software as a service trend is that in an increasing
number of cases, users don't have the software on their own devices,
but access a server, its location irrelevant, to get something done.
As a result, the customer base (of advertisers) is dramatically
smaller than the user base, delivering favorable sales force
Accordingly, software distribution channels are being completely
reinvented: the old goal used to be to get your product onto a shelf
and/or catalog page at Computer City, Egghead, or Micro Warehouse.
Note that all of those businesses are defunct, another indication of
deeper change in the industry. In a related development that sheds
further light on a complicated situation, PC Magazine subscriptions
have dropped from 6.1 million in 2003 to 4.8 million.
People Buy Features and Performance
There's a wonderful video that embodies this thinking perfectly: enter
"microsoft ipod" into the YouTube search bar. Microsoft apparently
produced this spoof internally, illustrating the trend toward "speeds
and feeds" in stark contrast to Apple's aura and powerful design
sense. Just run down the standard old-school software questions in
regard to Hotmail or Mapquest:
-What is the recommended processor?
-How much free disk space is required?
-What is the minimum memory required?
-How many transactions per second can the application handle?
-How fast can the application render/calculate/save/etc.?
The very mention of these former performance criteria in regard to the
most successful "applications" of our time highlights the
discontinuity between where we are and where we were. It's critically
important to note that the path from Lotus Organizer or the original
Encarta to Basecamp or Wikipedia involved a step-function change
rather than evolutionary progression.
Hire the Best Technical Team
There's no question that high-caliber architects and developers
matter. Look at the arms race among Microsoft, Amazon, Google, and
Yahoo to hire the giants of the industry: Gordon Bell, Brian Valentine
(see above), Adam Bosworth, and Larry Tesler, respectively, only begin
a very long list. But the outside-in dynamic of user-generated
content also allows such sites as del.icio.us or Grouper (now Crackle)
to thrive. In these kinds of businesses it's certainly imperative to
get top-flight operations and data-center professionals, no question,
but these folks are of a different breed compared to the breakthrough
innovators of the caliber mentioned above.
Quality is Built from the Inside Out
This area is tricky. Certainly the core application functionality and
engineering need to be built into the base architecture, as eBay
discovered a few years back. But no longer is the internal team the
only resource: many of the best businesses balance internal and
external talent, Amazon being exhibit A. In contrast, efforts built
on pure volunteer collaboration, such as the Chandler PIM and Mozilla
browser, have been outpaced by commercial ventures. It's also worth
reiterating that Apple runs a very closed shop very successfully: the
iPod and iPhone feel antithetical to the Web 2.0 mantra. It would
appear that in this regard, as in many others, several successful paths remain available.
Rows and Columns
While I don't want to oversimplify and assert that value has migrated
from nodes to links, the fact remains that the structure of business,
personal connections, and information is looking much more like a
spider web than a library card catalog. As scholarship from Rob Cross
at Virginia and others has illustrated, informal networks of personal
contacts, once exposed, often explain a corporation better than the
explicit titles and responsibilities. More recently, Mark Anderson at
Strategic News Service has connected some of the dots around Google
and Apple, at both the board level and elsewhere, contending that an
ecosystem is taking shape to challenge Microsoft. At the engineering
level, the very concept of social networking behind Twitter, flickr,
and the Dodgeball startup scooped up by Google represents a departure
from a conventional relational database mentality.
Calling this a trend would be premature, but the corporate
architectures at Microsoft, Google, and Apple mirror their varying
approaches to the market. Apple's share price includes a healthy
dose of respect for the management ability of Steve Jobs, in that
particular context, to both envision and execute. Conversely, the
achievement of Google, with the jury out on the model's staying power,
may lie in leadership's balancing of individual brilliance at
different layers of the hierarchy with financially realistic corporate
objectives. Finally, Microsoft appears to be working hard to define
an emerging management model as the founding generation hands off to
new COO Kevin Turner (from Wal-Mart) and CTO Ray Ozzie, long ago at
While it certainly includes a substantial element of buzzword-mania,
the shift from rows and columns to graphs -- whether in software
architecture (cf. Metaweb), business model (Facebook), or management
structure (Linux still matters here) -- merits watching for several
reasons. First, the combination of cheap and (remotely available)
processing, effectively infinite online storage, and functionality
tuned to these realities means that graphs are required to handle the
sheer scale of available data. Second, the ability to map and model
networks allows their structures to be better understood and utilized.
Finally, social groups get larger than could be managed in an
unconnected world -- according to a recent survey of 18,000 people
conducted by Nickelodeon, MTV, and Microsoft, "Globally, the average
young person connected to digital technology has 94 phone numbers in
his or her mobile phone, 78 people on a messenger buddy list and 86
people in his or her social networking community." This requires both
new ways to understand social connections and tools with which to
manage them. To underscore this shift, the North Carolina Attorney
General announced earlier this week that MySpace just ceased hosting
pages for 29,000 known sex offenders.
Taken together, these tendencies are reshaping the software business:
programming (as in putting content together) has joined programming
(as in coding) as a core competency for many kinds of businesses that
fall in the gaps between computing and media. The fusion also shakes
up conventional media, as we have noted earlier. The purely
push-based media model, used to advertise things primarily for largely
unmeasurable brand impact (unmeasurable at the level of the ad,
particularly), is being challenged by viewers and readers who want
more participation in both the experience (what used to be called
consumption) and the process (formerly known as publishing or content
creation). The YouTube-CNN debates feel to some extent like a
gimmick, but they appear to be a harbinger. As blogs, social
networks, and professional content get further jumbled, as Rupert
Murdoch seems to be intent on doing, the business models of media,
software, gaming, and transport will continue to feel the effects.