This article is made available to you by Boston University School of Management. Further posting,
copying or distributing is copyright infringement. To order more copies go to www.hbr.org.
Strategies for Two-Sided Markets
harvard business review • hbr.org
• october 2006 page 6
Same-side network effects.
Surprisingly,
sometimes it makes sense to deliberately ex-
clude some users from the network. Platform
providers normally welcome growth in the
user base on either side, because it encourages
growth on the other side. In addition to posi-
tive cross-side network effects, however, plat-
form managers must assess the possibility of
negative same-side network effects, which can
be quite strong. In most markets, sellers would
be happy to see fewer direct rivals; the same
can be true for buyers when goods are scarce.
For example, many auto parts manufacturers,
concerned about downward pricing pressure,
refused to participate in Covisint, a B2B ex-
change organized by auto manufacturers. Cov-
isint stalled, as did many other B2B market
makers that failed to recruit enough sellers. In
the face of strongly negative same-side net-
work effects, platform providers should con-
sider granting exclusive rights to a single user
in each transaction category—and extracting
high rent for this concession. The platform
manager then must make sure that sellers do
not abuse their monopoly positions; other-
wise, buyers will avoid the network. Online
car-buying services like Autobytel, which for-
wards consumers’ queries to a single dealer in
any given geographic territory, have suc-
ceeded with this strategy. Autobytel has
earned a modest profit over the past three
years; more to the point, it survived the dot-
com crash that extinguished many Internet
market makers with flawed strategies.
Users’ brand value.
All users of two-sided
networks are not created equal. The participa-
tion of “marquee users” can be especially im-
portant for attracting participants to the other
side of the network. Marquee users may be
exceptionally big buyers, like the U.S. govern-
ment. Or they may be high profile suppliers,
like anchor stores in malls. A platform pro-
vider can accelerate its growth if it can secure
the exclusive participation of marquee users in
the form of a commitment from them not to
join rival platforms. For many years, this kind
of exclusive arrangement was at the core of
Visa’s marketing campaigns (“…and they don’t
take American Express”). Of course, it can be
expensive—especially for small platforms—to
convince marquee users to forfeit opportuni-
ties in other networks. When the participation
of a few large users is crucial for mobilizing a
network, conflict over the division of value
Similar Networks, Different Pricing
On first inspection, PC and video game net-
works look similar. In both cases, end users
on one side wishing to link to software or
games on the other side buy a platform con-
sisting of an operating system (OS) bundled
with hardware—a PC or a game console. The
two businesses exhibit similarly positive
cross-side network effects: End users favor
platforms that offer a wide variety of comple-
ments. Developers favor platforms with more
end users because this improves the odds
that they will recover the fixed, upfront costs
of creating complements.
Notwithstanding these similarities, the PC
and game industries use very different pricing
models. In video games, end users are subsi-
dized. Platform providers like Sony PlaySta-
tion and Microsoft Xbox historically have
priced consoles at or below cost. Game devel-
opers are on the network’s money side; they
pay a royalty to console manufacturers of as
much as 20% of a game’s retail price. In the PC
industry, the money side and subsidy side are
reversed. End users are the money side, pay-
ing well above cost for the platform’s essential
element—its OS—which comes bundled with
PCs offered by OEMs like Gateway. Applica-
tion developers are the subsidy side. They pay
no royalties and receive free software develop-
ment kits from the OS vendors.
Why do these similar two-sided networks
have fundamentally different pricing struc-
tures? Video game consoles users—typically
teenagers—are both far more price sensitive
and quality conscious than typical PC users.
On average, each console owner buys just
eight games, which cost about $50 apiece.
Over the two- to three-year life of a console,
these precious titles are consumed sequen-
tially in intense bursts; gamers spend a great
deal of time—40 to 100 hours—with each title.
To deliver compelling quality, game devel-
opers incur enormous fixed costs. To amortize
these costs, they must be assured that the con-
sole has many users: Hence the need for a con-
sumer subsidy. Console providers police qual-
ity by imposing strict licensing terms and
charging a high royalty. This “tax,” absorbed by
the developers, helps weed out games of mar-
ginal quality. Developers cannot afford to offer
titles with weak sales prospects, once the tax is
added to their price.
By contrast, PCs are often purchased for
work and are otherwise more likely viewed
as household necessities than game con-
soles are, so price sensitivity is lower. Over
their lives, PCs accumulate scores of applica-
tions, ranging from the indispensable (such
as word processing) to the disposable (for
example, some casual games). Accordingly,
we observe a huge range of price and qual-
ity levels for applications.
It’s true that both PC users and gamers
value variety and quality and that develop-
ers in both networks value the ability to
reach a large installed base. However, gam-
ers’ need for quality seems to be stronger, as
does game developers’ need for large num-
bers of consumers.