Platform-Mediated Networks

Module Overview

November 8, 2007

2

Notes for Instructors• These slides are available for download as a file named “PMN

Slides.ppt” at http://www.people.hbs.edu/teisenmann/• I’ve provided comments on many of the slides, visible as “Notes”

(from the “View” menu, select “Notes Page”).• Feel free to adapt the slides in any way you see fit. In the spirit of open

source, if you make changes or additions, please forward a copy to me at teisenmann@hbs.edu so I can post them for others. I’m also happy to answer questions about the slides by email.

• The slides present concepts covered in an introductory module on platform-mediated networks, designed for use in existing courses on strategy, technology management, marketing, entrepreneurship, or ecommerce. The module is described in “Platform-Mediated Networks: Module Note for Instructors,” HBS #807-067.

• For instructors who wish to delve deeper, the module’s materials are drawn from a case-based MBA elective course, described in “Managing Networked Businesses: Course Overview for Educators” HBS #807-104.

3

Topics

1. Overview, Definitions, and Core Concepts

2. Network Mobilization

3. Platform Control

4. Platform Envelopment

4

Our vision for the last 20 years can be summarized in a succinct way. We saw that exponential improvements in computer capabilities would make great software quite valuable… In the next 20 years the improvement in computer power will be outpaced by the exponential improvements in communications networks. The combination of these elements will have a fundamental impact on work, learning and play.

Bill GatesThe Internet Tidal Wave, May 1995

5

1995

• Netscape IPO

• Rob Glaser launches RealNetworks

• VocalTec introduces VoIP

• First cable modems

• FCC auctions digital mobile spectrum

• Palm launches Pilot

6

1999

• IBM embraces Linux

• XML is created, paving the way for “web services”

• 802.11b (Wi-Fi) takes off

• Napster launched

• TiVO launched

7

2004: Explosive Growth in…

• Google’s valuation

• Skype

• RSS, podcasting

• Blogs

• Open source software

• Wikipedia

• BitTorrent

8

Platform-Mediated Networks

• Network users access a common platform that facilitates their interactions

• Platforms = subset of components and rules employed by users in most of their transactions– Components = hardware, software, services– Rules = technical standards, protocols for information

exchange, policies, and contracts that govern transactions

• Users rely on a platform when doing so is more efficient than unmediated bilateral dealings

9

User A User B

Network

User A User B

Components- Hardware- Software- Services

Rules- Standards- Protocols- Policies- Contracts

Architecture

Platform

10

eBay’s Platform

• Components– Browser and Internet access (from 3rd parties)– Website design – Bid tracking software– Shipping services (from 3rd parties)– Links to PayPal (an eBay-owned platform)– Etc.

• Rules– Registration requirements– Dispute resolution processes– Feedback system– Policies for bidding– Etc.

11

Network Effects• Definition: Network’s value to a user depends on the

number of other network users– “Value” = willingness-to-pay for network participation =

WTP for platform affiliation = cap on platform fees

• Properties– Can be negative, e.g., due to congestion – WTP tends to increase as “S”-shaped (logistic) function of

network growth – Network effects are demand-side economies of scale, i.e.,

they impact revenues• Network effects are not supply-side scale economies, which improve unit margins

through fixed-cost leverage• However, many networked businesses do enjoy strong supply-side economies

12

Platform-mediated networks do not necessarily encompass:

• Physical networks (e.g., gas utilities)• Social networks (e.g., your golf buddies)• Supply networks, aka “ecosystems,”

comprised of large firms and their most committed suppliers and customers (e.g., Wal-Mart, Toyota)

13

Why Study Platform-Mediated Networks?

• Large and growing share of global economy

• Distinctive management challenges

14

Large and Growing Share of Global Economy

• Not just digital industries, also:– Financial services, e.g., ATMs, credit cards, stock exchanges– Transportation, e.g., package delivery, airlines, travel agents,

reservation systems, fuel cell-powered cars– Retail, e.g., shopping centers, bar codes/RFID– Energy, e.g., grid + appliances, energy trading– Real estate, e.g., home buying– Health care, e.g., HMOs– Enterprise administration, e.g., recruiting, B2B procurement

• 60 of the world’s 100 largest companies earn most of their revenue from platform-mediated networks

15

Distinctive Challenges

• Concepts are new: first academic papers on “two-sided” networks published in 2003

• Platform-mediated networks are very complex, but oversimplification is common– Over- or underestimate strength of network effects – Price to network’s sides as if they were separate markets

• Errors can be fatal– Yahoo, Amazon failed in U.S. auctions– eBay exited Japan and China; failed to dislodge PayPal– NASDAQ collusion = $1 billion fine; ECN entry– IBM spent $1+ billion on OS/2– Apple forfeited Microsoft’s position

16

Distinctive Challenges

• Business Model Design

• Winner-Take-All Dynamics

17

Business Model Design

“This is like the Cambrian explosion 550 million years ago, when multicelled life first appeared on the scene. It was the greatest speciation ever seen, but it was also — which people forget — the greatest rate of extinction ever seen. We’re going to see all kinds of ideas tried, and the majority of them are probably going to fail.”

Jeff Bezos, CEO Amazon.com

(Business Week, 9/13/99)

18

19

Designing Business Models is Difficult Due To:

• “Two-sidedness” of most networks

• Bifurcation of platform roles

• Rapid growth, precluding trial-and-error

20

One- and Two-Sided Networks

• “One-sided” networks: Transaction partners alternate roles, e.g., e-mailers send & receive, traders buy & sell

• “Two-sided” networks: Users are permanent members of one distinct group — a “side” — which transacts with a second group, e.g., – Job seekers + recruiters– Card holders + merchants

21

A two-sided network has four network effects

• A same-side effect for each side, i.e., preference regarding number of other users on own side

• A cross-side effect in each direction, i.e., preference regarding number of users on other side

• Each effect can be positive or negative

Side 1

Platform

Side 2

22

Platform Roles: Sponsors and Providers

• Providers are users’ primary point of contact with platform

• Sponsors do not deal directly with users; rather, sponsors hold property rights that determine:

– Who may change platform technology

– Who may participate in network as a platform provider or network user

23

CARDHOLDERS

MERCHANTS GAMER DEVELOPER

ISSUINGBANKS

ACQUIRINGBANKS

VISA INTERNATIONAL

MICROSOFTXBOX

CONSOLE SDK

VISA =JOINTLY SPONSOREDSHARED PLATFORM

XBOX =SOLE SPONSORED

PROPRIETARY PLATFORM

24

Distinctive Challenges

• Business Model Design

• Winner-Take-All Dynamics

25

Schumpeterian Competition: Serial WTA Battles

• VCR DVD + PVR + VOD• LP CD download and/or subscription • Human market makers ECNs• 1G 2G 3G• Analog POTS VoIP• Broadcast cable IP TV• Atari Nintendo Playstation Xbox?• Wi-Fi WiMax?• Gasoline-powered cars Fuel cell-powered cars?• Barcodes RFID?

26

Blu-Ray vs. HD-DVD

27

Winner-Take-All

• Network effects increasing returns

winner-take-all, i.e., one platform prevails• Examples: Windows, eBay, PDF, DVD, fax,

real estate MLS• Winner-take-all implies loser takes nothing!• So, should you share platform with rivals?• If not, should you race to acquire network

users?

28

WTA Implications

• Sharing and racing are “bet-the-company” decisions, so organizational design is crucial

• Growth opportunities capital market bubbles

• WTA monopoly power government intervention

29

Platform Structure

• Winner-take-all: one platform serves the mature networked market

• Mono-homing: most users on a given side affiliate with a single platform

• Multi-homing: most users on a given side affiliate with multiple platforms

30

Platform Structure Examples

NetworkSide 1Users

Side 1Structure

Side 2Structure

Side 2Users

Fax Sender/receivers WTA (Homogeneous network)DVDs Consumers WTA Movie studiosMobile FeliCa Cell phone users WTA Retail stores, etc.Real estate MLS Home buyers WTA Home sellersWindows desktop O/S PC users WTA Application providersOnline auctions Buyers WTA SellersPDF Readers WTA Document creatorsInstant messaging Sender/receivers Mixed-mode (Homogeneous networks)Credit cards Consumers Mixed-mode Mixed-mode MerchantsYellow Pages Consumers Mixed-mode Mixed-mode MarketersRecruitment sites Job seekers Mixed-mode Mixed-mode RecruitersVideo games Consumers Mono- Mixed-mode Game developersShopping mall Consumers Mono- Multi- StoresSubscription music Consumers Mono- Multi- Artists/labelsMultichannel TV (DBSvs. cable)

Consumers Mono- Multi- Programming networks

31

Networked Market is More Likely to be

Served by a Single Platform When:

• The platform is a natural monopoly

OR… • Multi-homing costs are high AND • Network effects are positive and strong AND • Demand for differentiated features is weak

– OR dominant platform can offer such features selectively to users willing to pay premium

32

Natural monopoly: minimum efficient scale of production > mature market size

Figure 2: Unit Cost Curves Showing Constant vs. Increasing Returns To Scale

Quantity

Long-Run Unit Cost

Constant Returns ReturnRReturns RReturns

Quantity

Long-Run Unit Cost

Increasing Returns

33

Natural monopolies are rare, but evident in some networked businesses

• Past: local utilities (phone, power, cable TV), railways, postal delivery

• Present– Internet content delivery networks (e.g.,

Akamai, which has 85% market share based on huge investment in servers)

– Failed LEO satellite communications (e.g., Iridium, Teledesic)

34

Strength of Network Effects• Network effects are stronger when:

– Users demand novelty from repeated transactions (e.g., DVDs)

– Mobile users require geographic coverage (e.g., ATMs, Wi-Fi, refueling)

– Participants in a matching network have idiosyncratic needs and offers (e.g., home buying, executive recruiting)

• Access to network users is not valued equally– Some are extremely valuable (e.g., friends, family)

– Others are worth little/nothing (e.g., strangers in another country)

– Some have negative value (e.g., telemarketers, stalkers)

• Strong preference for variety yields“long tail”

35

Measuring Network Effects: Conjoint Analysis

• Interviews elicit reactions to product attributes—including network size—usually through paired comparisons – Hence, CONsidered JOINTly

• Requires consumers to understand attributes, so may not be reliable for breakthrough products

• Studies should be designed by market research professionals and cost between $50k-$250k

36

Homing costs: costs/investments incurred by user due to platform affiliation

Upfront •Search and negotiation

•Account setup, e.g., software configuration

•Initial hardware & software investment; system integration

•Training

Ongoing •Membership and transaction fees

•Maintenance costs; customer service hassles

•Tenure- or volume-based benefits

Exit •Account termination hassles and costs, e.g., changing email address, moving funds between brokerage accounts

•Contract severance penalties

•Salvage value of hardware, software

37

Switching Costs

• Out-of-pocket expenses and inconveniences incurred by network users (or by platform providers on their behalf) when users switch from one platform to another

• Network effects may deter or encourage switching, depending on the relative sizes of rival platforms.

• However, network effects and switching costs are conceptually distinct and should not be confused

38

Homing vs. Switching Costs

2 SETUPS + 2 ONGOING

1 SETUP + 1ONGOING

Mono-homing

Switching

Multi-homing

2 SETUPS + 1 TERMINATION + 1 ONGOING

39

Homing vs. Switching Costs

• Usually they move together, e.g., satellite radio (both high), package delivery (both low)

• Sometimes, switching cost is high but multi-homing cost is low– eBook readers: costly to switch because you must replace entire

library, but not very costly to multi-home

– Email account: costly to switch because you must notify all your contacts, but not very costly to multi-home

40

Homing vs. Switching Costs

• When predicting whether a new networked market will be served by a single platform, focus on multi-homing costs

• When predicting whether to race to acquire network users, focus on switching costs

• When predicting whether an established platform is vulnerable to displacement by a new platform, focus on both multi-homing costs and switching costs

41

Users’ Preferences for Differentiated Platform Functionality

• User segments have different needs– DBS picture is sharper than cable’s

• Appealing for sports and movie lovers?

– DBS requires a set-top box for every TV • A liability for large families?

• Can functionality be offered selectively to users willing to pay a premium?– To match DBS picture quality, cable would

have to convert to all-digital architecture, so selectivity is not an option

42

Homing Logic: Affiliate With Platform When ∑VP + I ≥ H

• VP = expected value from ability to interact with a potential transaction partner

• I = value from platform, independent of platform size

• H = homing cost

43

Multi-Homing Logic

• For platforms A and B, calculate ∑VP + I ≥ H

• Consider incremental value from access to potential transaction partners from multi-homing on A and B, compared to mono-homing with the best single platform, A– B may offer access to some partners not affiliated with A

– Due to differentiation, B may offer more value for certain transactions than A

• Compare incremental value to incremental homing cost

44

WTA Potential?DVD: WTA

Credit Cards: Multi-Homing

Strength of Network Effect

High for most users on both sides, i.e., consumers and studios

High for most users on both sides, i.e., card holders and merchants

Multi-Homing Cost

High for both sides Low for both sides

Demand for Inimitable Features

Low due to technical standardization of TV

High: “revolve” vs. charge (i.e., pay-in-full with no preset limit)

45

Mono-Homing + Multi-homingExample: Online Subscription Music

Side 1 Mono-Homing: Consumers

Side 2 Multi-Homing: Music Companies

Strength of Network Effect

•High: want access to all music

•High: revenue increases in direct proportion with user base

Multi-Homing Cost

•High: monthly fees, playlist management

•Low: duplicated legal work, but zero inventory and modest incremental production costs due to digital distribution

Demand for Inimitable Features

•Moderate: differences include quality of editorial content, IM integration

•Low: music companies have similar needs for DRM, promotional support, etc.

46

Topics

1. Overview, Definitions, and Core Concepts

2. Network Mobilization

3. Platform Control

4. Platform Envelopment

47

Barriers to Network User Adoption• Business stealing risk: too many rivals on your side

– Unlike other risks, does not involve provider/user trust

• Holdup risk: dominant platform extracts too much value (e.g., Microsoft OS for cable set-top box or mobile phones)

• Stranding risk: 1) user backs wrong side in WTA battle; 2) platform fails to invest; 3) no backward compatibility

• Integration risk: platform provider integrates into network user role (e.g., Microsoft with Office & Halo)

• Favoritism risk: to attract users, extract rent, or serve self-interest, platform provider skews terms in one side’s favor (e.g., Covisint)

• Relationship risk: In 2-sided net, who controls access to users on other side? (e.g., Xbox vs. EA; Autobytel loans)

48

Willingness-To-Pay and Network Size• D1 is traditional demand curve if

customers/prospects expect a network size equal to N1

• If customers/prospects expected a bigger network, N2, then demand curve would shift outward to D2

– Bigger network is more useful

– Existing customers will consume more

– New prospects will join network

• Outward shifts become smaller as expected network size nears saturation point, Ns, due to

– Later adopters not as valuable as early adopters

– Congestion

WTP

Network Size

N1 N2 N3

P2

P3

P1

D1

D2 D3

Ns

49

Beware Metcalfe’s Law• According to Metcalfe’s Law, the value of a network grows

with the square of growth in its user base– In a mesh network with each of N nodes connected to every other node,

there are N x (N-1) links, i.e., ~ N2

• But, ∆ Value = ∆ N2 only if:– Per capita transaction volumes also increase by ∆ N2

– Each user’s WTP per transaction remains constant as her transaction volume grows

• In the real world: diminishing returns set in and constraints on attention/budget prevail

50

Upward Sloping Demand in Network Industries

• “Fulfilled expectations” curve traces price that would exactly fulfill demand, for a given expected network size

• Curve is upward sloping in domain where network effects dominate price elasticity effects

WTP

Network Size

N1 N2 N3

P2

P3

P1

D1

D2D3

Fulfilled Expectations Demand Curve

51

Multiple Equilibria in Network Industries

• Assume competitive market with Price = Marginal Cost

• Market has three equilibrium points: N0, N1, and N2

• N1 is not a stable equilibrium– If expected network size < N1,

firms cannot make money, so N0 prevails

– If expected network size is > N1, supply will surge to N2

• Implication: network markets are tippy —they can stall, and they can grow explosively— and expectations matter!

WTP

Network Size

N1 N2

Fulfilled Expectations Demand Curve

Marginal Cost

P

N0

52

The Penguin Problem

53

The Penguin Problem(Farrell & Saloner, 1986)

• In new markets, uncertainty is high; individuals have different expectations about long-term demand

• If user base is fragmented, users cannot communicate expectations or coordinate behavior. Network may stall, even with strong network effects!

• No-one moves unless everyone moves, so no-one moves, like penguins afraid to be first to dive for food

54

Two ways to categorize platforms:

• Structure: Proprietary, Shared, Joint Venture, Licensor

• Function: Connectivity, Variety, Matching, Price Setting

55

Categorization Based on Platform Function

• Connectivity: platform facilitates point-to-point exchange of information, physical goods, or passengers (e.g., fax; Fedex)– More points connected more valuable network more points etc.

• Variety: platform elicits supply of a variety of complements from one network side and promotes complement consumption by the other side (e.g., console + games; credit card + merchant locations)– More complements more demand-side users more complements etc.

• Matching network: platform matches transaction partners with heterogeneous needs and offers (e.g., real estate; online dating)– More buyers more sellers improved odds of suitable transaction more

buyers etc.

• Price-setting network: platform facilitates exchange of well-defined items valued differently by trading partners (e.g., stock exchange, gambling book-maker)– More buyers more sellers reduced price volatility more buyers etc.

56

Categorization Based on Platform FunctionPartner’s Identity

Generally Known in AdvancePartner’s Identity Generally Discovered

Through Transaction

One-Sided

Connectivity

Phone, fax, IM, SMS, email, Ethernet, etc. Package/freight delivery Passenger air and rail lines Secure document delivery (e.g., Tumbleweed) Online invitations (e.g., eVite)

Price-Setting

Securities and commodity exchanges Online /offline auctions Wholesale electric power markets Gambling bookmakers

Multi-Sided

Variety

DVDs, VCRs, HDTV, CD audio, video gameconsoles, satellite radio, etc.

World Wide Web Internet content delivery (e.g., Akamai) Internet-based media players (e.g., PDF) PC, smart phone, PDA operating systems Media aggregators (e.g., YouTube) Ad-supported media (e.g., TV networks) Wi-Fi: sides = end users & access points Barcodes or RFID for product ID Credit cards, email payments (e.g., PayPal) Fuel cell-powered cars & refueling stations ATMs Shopping malls HMOs Electric grid: sides = outlets & appliances

Matching

Real estate brokerage Online and offline travel agencies Online car buying Online and offline insurance brokerage Mutual fund “supermarkets” (e.g., Schwab) Business-to-business online exchanges Executive search, online recruitment Yellow Pages, paid search (e.g., Google) Nightclubs, online personal ads and dating

services, marriage brokers, brothels Markets for expertise (e.g., Yahoo Answers)

57

Key Challenges in Network Mobilization

Connectivity•Ensure reliability•Optimize reach (infrastructure investment to extend network vs. incremental revenue)

Price Setting•Balance users’ different needs for:

–Anonymity–Transparency with respect to most recent price and best current offer–Transaction speed–Flexibility to trade with preferred partners

Variety•Optimize price structure, exclusivity•Elicit complement supply despite:

–Risky platform-specific investments–Conflict with large users over rent division, end user relationship control–Concerns about platform provider integration into user role

•Help end users find complements

Matching•Optimize price structure, exclusivity•Build user trust•Address worries about business stealing•Manage concerns about conflicts of interest, ownership of data, etc.•Avoid disintermediation, off-exchange trading

58

Internalizing Externalities• Externality = benefit or harm experienced by B due to A’s

actions, with no compensating payment – Positive externality: I admire your flowers – Negative externality: your pig pen is upwind

• Network effects are externalities• Three approaches for internalizing externalities

– #1: Side payments between users, e.g., I buy grandma’s videophone– #2: In two-sided network, permanent subsidization of one side– #3: Inter-temporal internalization: sponsor “primes the pump” with

discounts to early adopters; this leads others to revise expectations/WTP

• For #2 and #3, proprietary platform is typically required; otherwise, free riders exploit subsidy – Exception: sponsors can dictate pricing terms that underwrite subsidies, as

with credit cards and real estate

59

Strategies for Network Mobilization

• Pricing – Permanent: should one side be subsidized?– Penetration: should platform providers race to

acquire network users?

• Participation– Exclusivity: should platform providers seek

exclusive relationships with some users?– Integration: should platform providers become

network users to avoid chicken & egg problem?

60

61

Pricing Puzzles

• What explains “Ladies Night”?

• Why do similar industries charge different sides?– Video games: end user subsidized; developer

pays– PCs: end user pays; developer subsidized

62

Assume monopolist serves markets A and C

Marginal cost = 0

With no network effect between A and C, monopolist prices each product to maximize P x Q

PA

P*A

Q*A QA

PC

P*C

Q*C QC

Case 1: No Network Effect

PA

P*A

Q*A QA

PA

P*A

Q*A QA

PC

P*C

Q*C QC

PC

P*C

Q*C QC

Case 1: No Network Effect

Market A Market C

63

Case 2: Positive Cross-Side Network Effects

Compare ((P^A x Q^A) – (P*A x Q*A) – (P*B x Q*B)) to ((P^B x Q^B) – (P*B x Q*B) – (P*A x Q*A))

Demand on each side grows if QMAX is sold on the other side at a zero price. Charge the side that most highly values growth in the other side.

Side A Side B

P^A

P*A

Q*A Q^A

P^B

P*B

Q^BQ*BQMAXA

QMAXB

64

Strategies for Network Mobilization

• Permanent Subsidies

• Accelerated Growth Strategies

• Exclusivity

• Vertical Integration

65

66

67

Racing to Acquire Users

• Firms race to acquire customers when they face:– Increasing returns to scale

– High switching costs

• Late mover advantages—e.g., reverse engineering, free riding on pioneer’s missionary marketing—may discourage preemptive investment

• Will investors fund racing strategies?

68

Racing to Exploit Increasing Returns

• Without increasing returns:– Maximize: Profit = Revenue - Cost– Implies: Marginal Revenue (MR) - Marginal Cost (MC) = 0– MR = MC

• With increasing returns due to network effects:– Willingness-to-pay (WTP) = f(Number of other users)– So, adding a user boosts every other users’ WTP – MR + NPV of extra revenue from other users = MC– To maximize profit, you should price lower (reducing MR)

or invest more in marketing (increasing MC)– In other words, you should race to acquire customers!

69

High switching costs also encourage firms to race to acquire customers

• In an otherwise competitive market, a firm should earn profit equal to the sum of switching costs confronting its customers– $100 price just covers costs of producing/delivering one unit– Assume consumers buy one unit of product in periods 1 and 2– If switching costs are $50, then firm A can price at $149.99 to its

existing customers in period 2. – To steal A’s customers, Firm B must fully compensate customers for

switching costs – However, $150 is the best price B can offer and still recover its cost

• These economics will fuel a race to acquire unaffiliated customers in period 1: each firm will price at $50.01

70

The Payoff from Racing to Acquire Customers

• As investment rises, NPV of a new customer eventually declines due to:– Broadening beyond natural

segments– Prematurely soliciting prospects– Retaliatory responses– Scalability constraints

• I* corresponds to Lifetime Value of Customer = 0 for the last customer you acquired– Average LVC should be positive

Current Period Investment inCustomer Acquisition Efforts

Impacton NPV

Zero

I*

71

Lifetime Value of a Customer

• Rule: acquire customers until Lifetime Value of a Customer (LVC) = 0 for last customer acquired, i.e., PV of variable contribution margin = customer acquisition cost (including incremental capital expenditures)– LVC does not allocate fixed costs. If LCV is positive, but you don’t have enough

customers to cover the present value of fixed costs, you’ll never earn a profit!– LVC provides a guideline for marketing spending; it is not a substitute for

detailed cash flow analysis

• Issues with LVC in networked industries– Issue #1: LVC should include new customer’s impact on other customers’

willingness to pay – Issue #2: Harder to calculate LVC in two-sided networks (e.g., Monster.com)– Issue #3: With some connectivity networks (e.g., Federal Express), variable

costs depend on user density, so you must project customer base to calculate LVC

72

What might explain departures from long-term value maximizing investment levels?

• Investment < I*– Resource allocation processes

– Funding constraints

– Antitrust constraints

• Investment > I*– Inadvertent: Overconfidence

– Deliberate: Exploit overvaluation

Current Period Investment inCustomer Acquisition Efforts

Impacton NPV

Zero

I*

73

Managing through a speculative valuation cycle

• Can/should managers exploit overvaluation? – Currency for acquisitions (Google)– Buffer (Akamai)

• How will different types of investors react to bust?– Private vs. public equity (e.g., Fedex vs. UPS)– VC vs. strategic investor– Family (e.g., Cox)

74

Strategies for Network Mobilization

• Permanent Subsidies

• Accelerated Growth Strategies

• Exclusivity

• Vertical Integration

75

Topics

1. Definitions and Core Concepts

2. Network Mobilization

3. Platform Control

4. Platform Envelopment

76

Platform Roles: Sponsors and Providers

• Providers are users’ primary point of contact with platform

• Sponsors do not deal directly with users; rather, sponsors hold property rights that determine: – Who may change platform technology – Who may participate in network as a platform

provider or network user

• Each role may be filled by one firm or many; sometimes, a single company fills both roles

77

Platform Types

One Provider Many Providers

One Sponsor

Proprietary

•Macintosh

•Monster.com

•Playstation

Licensor

•Windows

•American Express

•Scientific-Atlanta

Many Sponsors

Joint Venture

•CareerBuilder

•Orbitz

•Covisint

Shared

•Linux

•Visa

•Real Estate MLS

78

Key Design Decision

• Preserve proprietary control of new platform, e.g., eBay, Federal Express, Google?

Or…

• Share platform with rivals who offer compatible but differentiated platform products and services, e.g., barcodes, Wi-Fi, DVD?

79

Outcomes Vary

• Proprietary platforms may prevail– Akamai vs. Content Bridge– eBay vs. FairMarket

• Shared platforms may prevail– Citibank vs. Cirrus ATM networks

• Proprietary and shared platforms may coexist– Macintosh and Linux– American Express (until 2004) and Visa

80

Decision Rules

• Aspiring platform providers with a big edge (e.g., via strong patents) should almost always pursue proprietary approach

• When many evenly matched firms can simultaneously launch a new platform:– Winner-take-all propensity encourages shared approach

– Free-rider problems encourage proprietary approach

81

If a Proprietary Platform Serves the Entire Market…

• Users will worry about hold-up by monopolist

• Prospective providers will worry about losing 100% of their investment if they are not the winner

82

Free-Rider Problems with Big Upfront Investments

User Subsidization Centralized Infrastructure

83

WTA Potential?

Yes No

Free Rider Issue?

Yes

Proprietary Favored

•Proprietary examples: PayPal, Yellow Pages in smaller cities

•Shared examples: WWW, real estate MLS

Proprietary Favored

•Proprietary examples: video games, paid search

•Coexistence examples: NYSE/ECNs; credit cards

No

Shared Favored

•Shared examples: DVD, fax, barcodes, Wi-Fi, SMS

Coexistence Common

•Coexistence examples: Symbian + Blackberry; Linux + Mac

84

Competing on a Shared Platform

85

Competing on a Shared Platform

• Pulled together by positive impact on platform adoption and each firm’s survival odds– Fear of “penguin effect”– Fear of WTA outcome (e.g., mobile FeliCa)– Fear of Balkanization (e.g., pre-standard 11n)– Fear of outside threats (e.g., mobile phone OEMs vs. Microsoft)

• Pulled apart by reduction in appropriability of platform rent – Concern about too many rivals– Desire to incorporate own technology in standard (802.11)– Desire to differentiate a standardized product (e.g., J2EE; 802.11)– Friction between IP owners and free riders (e.g., RAND-Z)– Propensity to form coalitions that exclude close rivals (e.g., KDDI +

JCB vs. DoCoMo + Sumitomo)

86

Value Capture Strategies for Shared Platform Providers

• Restrict Membership– Can backfire if excluded parties sponsor competing platform, e.g., Sun vs.

IBM/Microsoft in web services

• Profit from IP– If too low, IP-rich firms may avoid platform

– If too high, IP-poor may avoid platform and stalemates may occur as firms vie to contribute technology

• Profit from Implementation– Freely contribute IP to secure time-to-market advantage or develop

proprietary extensions to industry standards

– But, proprietary extensions may splinter a platform, e.g., Unix

87

Key Skills

• Timing: Too soon = ignored; too late = blocked • IP Management: Avoid “submarine” patents and patent “thickets” via:

– Standard-setting organization (SSO) disclosure rules– Patent pools– Patent hoarding to ensure “mutual assured destruction”

• Diplomacy: Tactics include appeasement, intelligence gathering, logrolling, bluffing, burning bridges, irrational commitment to dogma, etc.– SSO delegates require rare mix of technical prowess and diplomacy skills; their

loyalties are often divided between profession and firm

• Organizational Design: Ad hoc vs. established SSOs; “forum shopping”; SIGs; etc.

88

Topics

1. Definitions and Core Concepts

2. Network Mobilization

3. Platform Control

4. Platform Envelopment

89

Platform Envelopment• Platform Envelopment: Entry by one platform provider

into another’s market, combining its own functionality with the target’s in a multi-platform bundle that leverages common components and/or shared user relationships

• Envelopment is a powerful force shaping the evolution of platform-mediated networks– Path to platform leadership change that does not require

Schumpeterian creative destruction– Driver of industry convergence– Strategy for cross-layer competition

90

Platform Entrenchment

• Single platform often dominates due to demand- and supply-side scale economies, e.g., Windows, eBay, PDF

• Dominant platforms often difficult to displace due to network effects and switching costs To succeed, entrants generally must offer revolutionary products, e.g., ECNs, fuel-cell cars, IPTV, VoIP, WiMax

• Hence, platform leadership change often entails serial winner-take-all battles, with superior new platforms displacing old ones

• Platform envelopment is a second path to platform leadership change that does not require Schumpeterian creative destruction

91

Platform providers serving separate markets may share users and components

Platform A

Platform B

Platform A

Platform B

Users Users

92

Platform Envelopment

• B enters A’s market with a multi-platform bundle

• A is vulnerable if it cannot match an appealing bundle

Platform A

Platform B

93

Envelopment in Subscription Music• Prior to 1999, RealPlayer dominates

streaming media

• Microsoft’s WMP envelops RealPlayer, bundling streaming server into NT at no extra cost

• Real launches subscription music service Rhapsody

• Yahoo discounts subscription music, cross-selling via portal

• Apple could bundle subscription service into iTunes, leveraging iPod

• AT&T Wireless bundles Yahoo subscription music with cell phone service

Labels Consumers

Real

AT&T

Apple

Rhapsody

Yahoo

Microsoft

94

Convergence Envelopment

95

Xbox 360 vs. PS3

96

Cable vs. Telco

97

Cross-Layer Competition Envelopment

PCs

Components

Applications

Operating Systems

PCs

Components

Operating Systems

Middleware

Applications

98

Telephone and Telegraph

“One newspaper account described AT&T, in its slow but steady envelopment of the telegraph industry, as a ‘quiet octopus.’”

Hochfelder, Business History Review,

2002

99

Mechanism: Bundling

• Efficiency Gains: Transaction and production cost savings; improved user experience

• Price Discrimination: Reduces heterogeneity in consumers’ aggregate valuations

• Strategic Advantages: Extend market power into complement market; protect core market by foreclosing rivals’ access to key complement

100

•Monster vs. LinkedIn

•UPS vs. Federal Express

•Blockbuster vs. Netflix

•Cable vs. Telco

•DoCoMo vs. JCB

•iPhone vs. Blackberry

•Microsoft vs. Real

•eBay vs. PayPal

•Microsoft vs. Adobe

Examples

Neutralize emerging threat

LimitedExtend market power, defend core

Strategic Advantages

Moderate to strongLimitedPrice Discrimination Benefits

•Production cost synergies

•Some economies in marketing

•Only viable with strong production cost synergies

•Economies in marketing

•Avoid double marginalization

Efficiency Gains

Weak substitutesFunctionally unrelatedComplementsRelationship

IntermodalConglomeration

(Convergence)

Foreclosure

(Cross Layer)

Limited

101

Defending Against Envelopment

• Match bundle, but bundle vs. bundle competition can be fierce

• Find a “big brother” via merger (e.g., Scientific-Atlanta + Cisco; Lotus + IBM) or alliance (e.g., RealNetworks + Comcast, Cingular)

• Cede and redeploy, e.g., RealNetworks shifts from streaming to subscription media

• Sue, e.g., Novell, RealNetworks, Netscape, Sun vs. Microsoft