Markdown Editor

MBA Platform Strategy Concept Card Template

Multi-sided platforms, cross-side network effects, two-sided pricing with Solver, subsidy side, critical mass, launch, openness, envelopment

MBA et écoles de commerceCompletRochet and Tirole (2003); Armstrong (2006); Katz and Shapiro (1985); Eisenmann, Parker and Van Alstyne (2011)platformstrategycardmath

Mode d'emploi: Use it when deciding whether to build, open or price a platform. Estimate which side values the other more first, then set the price structure before the level. The main pitfall is pricing each side as a separate product.

Aperçu

Platform Strategy Concept Card: [Platform or case name]

A concept card for platform strategy as it is taught in a second-year MBA technology strategy elective: how multi-sided platforms create and capture value, how cross-side network effects change pricing (which side to subsidise and by how much), why platform markets can tip, how to solve the launch (chicken-and-egg) problem, and how openness, governance, multi-homing and envelopment shape competition. It builds on the platforms and network effects topic of the strategy core (mba/mba-strategy-guide, section 6) and on mba/mba-growth-strategy-card, and links to them instead of repeating them. It is for MBA students preparing a case or exam and for managers deciding whether to build, open or join a platform. The card gives the idea first, then formulas with a notation table, a fully solved two-sided pricing example for a Northfield Outdoor Co. resale marketplace, a half-solved critical-mass example, an unsolved exercise on Corvane Robotics' fleet software, a self-test and a separate answer key. All companies and numbers are fictional. Source basis: Katz and Shapiro (1985), Rochet and Tirole (2003), Caillaud and Jullien (2003), Parker and Van Alstyne (2005), Armstrong (2006), Eisenmann, Parker and Van Alstyne (2011), Boudreau (2010), Zhu and Iansiti (2012), Hagiu and Wright (2015) and Parker, Van Alstyne and Choudary (2016), written up in original words and examples. This is study material, not legal or competition-law advice. Replace every [bracketed] value and every row marked Example.

Document control

FieldValue
Course or project[Technology strategy elective, session 4 / client project / product strategy]
Learner or team[Your name or learning team]
Platform analysed[Northfield Outdoor Co. resale marketplace (fictional) / your platform]
Sides of the platform[Buyers and sellers / sites and robot makers / users and developers]
Currency and unit[USD per user per year; users in thousands]
Evidence sources[Pilot data, surveys, conjoint, cohort data, competitor filings]
Reviewed by[Instructor, mentor or product leader]
DateYYYY-MM-DD
Version[0.1]

Contents

When to use it

One-sentence rule: use this card when a business creates value by enabling direct interactions between two or more distinct groups, and the value each group gets depends on how many (and which) members of the other group join.

Use it whenUse something else when
A firm considers turning a product into a platform (opening an interface, inviting third parties)The firm sells a product with no third-party participation (mba/mba-pricing-card)
You must price a marketplace, payment network, app store or software ecosystemYou need a one-sided price from an elasticity estimate (mba/mba-demand-elasticity-card)
A new platform cannot attract one side without the other (launch problem)You need a full launch and channel plan (marketing/go-to-market-plan)
Two platforms compete and users may join both (multi-homing)The question is industry structure without network effects (mba/mba-five-forces-card)
An incumbent platform could bundle your function into its own offer (envelopment)You need to value the platform business (mba/mba-dcf-valuation-card)

The idea in plain language

What a platform is. A multi-sided platform enables direct interactions between two or more distinct groups of participants, each of which chooses to affiliate with it (Hagiu and Wright 2015). Buyers and sellers on a marketplace, cardholders and merchants on a payment network, users and developers on an operating system, and warehouses and robot makers on an orchestration layer are examples. The defining features are that the sides keep control over the key terms of their interaction (price, product, effort) and that each side's value depends on participation by the other. A firm that buys inventory and resells it is a merchant, not a platform, even if it has many suppliers.

Network effects. Katz and Shapiro (1985) formalised network externalities: a user's benefit rises with the number of other users of a compatible product. Direct (same-side) effects arise when users value other users of their own kind (a messaging network, a member community). Indirect or cross-side effects arise when one side values the other side (more buyers attract sellers, more sellers attract buyers). Effects can also be negative: more sellers mean more competition for each seller, and too many advertisers annoy users. Network effects are not the same as economies of scale on the supply side: they raise willingness to pay, while scale lowers cost.

Price structure, not only price level. Rochet and Tirole (2003) showed that on a two-sided platform, the allocation of the total price between the two sides affects the volume of interactions, not just the total. A platform therefore chooses a price structure: often one side is charged little, nothing or less than cost (the subsidy side), and the other side pays (the money side). Armstrong (2006) gave the intuition for a monopoly platform: the price to a side equals its cost, minus the external benefit that one more member of this side brings to the other side, plus a markup that depends on how price-sensitive this side is. The side whose members are most valuable to the other side, and most price-sensitive themselves, gets the low price. Parker and Van Alstyne (2005) showed the same logic for information goods: a firm may give away a product to one market to increase profit in a complementary one.

Tipping and critical mass. Strong network effects create positive feedback: a platform that gets ahead becomes more attractive and pulls further ahead. With strong effects and low differentiation, the market can tip towards one platform (Katz and Shapiro 1994; Arthur 1989 on increasing returns and lock-in). Adoption models then have more than one equilibrium: a low one with few users and a high one with many, separated by a critical mass. Below it, users leave because others leave; above it, users join because others join. Expectations matter: users join the platform they expect others to join, so announcements, commitments and early visible traction are strategic tools.

The launch problem. Neither side wants to join an empty platform. Caillaud and Jullien (2003) modelled the "chicken and egg" problem and the divide-and-conquer strategy: subsidise one side to join, then charge the other side, which now finds the first side present. Common launch tactics are seeding (supplying one side yourself, as a marketplace might stock its own inventory first), piggybacking on an existing user base, a single-side value proposition (a tool that is useful before any network forms), micro-market launch (win one city, campus or industry niche before expanding) and marquee participants (sign a few large members whose presence attracts others).

Openness and governance. A platform owner decides how open to be. Boudreau (2010) distinguished granting access to complementors (letting third parties build on the platform) from devolving control (giving up control over the platform itself), and found that granting access was associated with much faster development of new devices in his handheld-computer data, while devolving control had a smaller effect. Openness raises variety and adoption but reduces the owner's ability to capture value and to keep quality high. Governance tools are the rules of access, pricing to complementors, quality certification, data sharing and dispute resolution. Cennamo and Santaló (2013) found that pursuing several competitive approaches at once (for example, exclusive content and broad variety) can hurt platform performance, because the approaches conflict.

Competition between platforms. Tipping is not inevitable. It is limited by multi-homing (users or complementors joining several platforms at low cost), by differentiation (platforms serving different needs), by congestion or negative same-side effects, and by regulation. Zhu and Iansiti (2012) showed that an entrant can overcome an incumbent's network advantage if its quality is high enough and consumers are not too forward-looking. Incumbents can also be attacked from adjacent markets: platform envelopment (Eisenmann, Parker and Van Alstyne 2011) is entry by bundling the target platform's function into a multi-platform offer that shares users with it. A platform that owns a large overlapping user base can envelop a standalone rival even if the rival's product is better.

Regulation. Large digital platforms now face specific rules in several jurisdictions. As a dated fact: as of October 2026, the EU Digital Markets Act (Regulation (EU) 2022/1925) has applied since 2 May 2023; the European Commission made its first gatekeeper designations on 6 September 2023 (six firms), and its DMA site lists seven gatekeepers and 23 designated core platform services. Under Article 3(2) a provider is presumed to be a gatekeeper for a core platform service when it has EU turnover of at least EUR 7.5bn in each of the last three financial years or an average market capitalisation of at least EUR 75bn in the last financial year, provides the service in at least three member states, and has at least 45 million monthly active end users and at least 10,000 yearly active business users in the EU, the user thresholds met in each of the last three financial years. The presumption can be rebutted, and the Commission can also designate firms below the thresholds after a market investigation. Note on sources: these figures were checked on reproductions of the Official Journal text and law-firm summaries, because EUR-Lex could not be opened at checking time; confirm them in Regulation (EU) 2022/1925 on EUR-Lex. Check the current list and obligations on the Commission's DMA site before you rely on them, and take any legal question (self-preferencing, data use, interoperability) to counsel.

Formulas and notation

Participation on each side (linear demand with cross-side effects, in the spirit of Armstrong 2006):

nB=vB+αB nS−pBtB,nS=vS+αS nB−pStSn_B = \frac{v_B + \alpha_B\, n_S - p_B}{t_B}, \qquad n_S = \frac{v_S + \alpha_S\, n_B - p_S}{t_S}

Solved for participation (requires tBtS>αBαSt_B t_S > \alpha_B \alpha_S for a stable interior solution):

nB=tS(vB−pB)+αB(vS−pS)tBtS−αBαS,nS=tB(vS−pS)+αS(vB−pB)tBtS−αBαSn_B = \frac{t_S (v_B - p_B) + \alpha_B (v_S - p_S)}{t_B t_S - \alpha_B \alpha_S}, \qquad n_S = \frac{t_B (v_S - p_S) + \alpha_S (v_B - p_B)}{t_B t_S - \alpha_B \alpha_S}

Platform profit:

Π=(pB−cB) nB+(pS−cS) nS\Pi = (p_B - c_B)\, n_B + (p_S - c_S)\, n_S

Profit-maximising prices (first-order conditions; concavity requires 4tBtS>(αB+αS)24 t_B t_S > (\alpha_B + \alpha_S)^2):

pB∗=cB−αS nS∗+tB nB∗,pS∗=cS−αB nB∗+tS nS∗p_B^{*} = c_B - \alpha_S\, n_S^{*} + t_B\, n_B^{*}, \qquad p_S^{*} = c_S - \alpha_B\, n_B^{*} + t_S\, n_S^{*}

Critical mass with a direct network effect (one community of NN potential members, standalone value uniform on [0,V][0, V], network value α\alpha per member, price pp, with Nα>VN\alpha > V and V<p≤NαV < p \le N\alpha):

n∗=N p−VNα−Vn^{*} = N \, \frac{p - V}{N\alpha - V}
SymbolMeaningUnit in the examples
nB,nSn_B, n_SParticipants on side B (buyers) and side S (sellers)Thousands of users
pB,pSp_B, p_SPrice charged to each side (fee or annual equivalent); negative means a subsidyUSD per user per year
vB,vSv_B, v_SStandalone value: what the most eager member gets with no one on the other sideUSD per user per year
αB\alpha_BValue to one buyer of one more thousand sellers (cross-side effect on B)USD per thousand sellers
αS\alpha_SValue to one seller of one more thousand buyers (cross-side effect on S)USD per thousand buyers
tB,tSt_B, t_SHeterogeneity: how fast willingness to join falls as more members join (inverse demand slope)USD per thousand users
cB,cSc_B, c_SCost to serve one member for a yearUSD per user per year
Π\PiPlatform profitUSD thousand per year
αSnS\alpha_S n_SExternal benefit one more buyer brings to all sellers (the subsidy term in pB∗p_B^{*})USD per year
tBnBt_B n_BMarket-power markup on buyersUSD per year
N,V,α,pN, V, \alpha, pCommunity size, top standalone value, direct network value per member, price (Worked example B)Thousands, USD
n∗n^{*}Critical mass: the unstable middle equilibriumThousands of members

Read the pricing formula as three terms: cost, minus the externality the member brings to the other side, plus the markup for this side's own demand. If the externality is larger than cost plus markup, the price is negative: the platform pays members of that side to join (free service plus rewards, credits or bonuses).

Assumptions and limits

  1. One platform, no rival. The pricing formula is for a monopoly platform. With competing platforms, the price on a side also depends on whether that side single-homes or multi-homes; Armstrong (2006) showed that platforms compete hard for single-homing sides and charge high prices to multi-homing sides that they can reach exclusively (competitive bottlenecks).
  2. Membership fees, not transaction fees. The examples price membership (an annual fee or its equivalent). Many platforms charge per transaction (take rates); Rochet and Tirole (2003) analyse transaction pricing, where the structure result also holds. Convert carefully: a take rate is a price per unit of value transacted.
  3. Linear demand and constant cross-side values. Real cross-side effects often flatten (the thousandth seller of used tents adds less than the tenth) and can turn negative (congestion, competition among sellers). Estimate α\alpha from data for the relevant range.
  4. Quantity, not quality. The model counts members. Users value relevant, high-quality members: one reliable seller of the item you need is worth more than many irrelevant ones. Curation and matching quality matter as much as size.
  5. Stability and expectations. The formulas describe an equilibrium. Whether the platform reaches it depends on expectations and on the launch path; the same prices can support an empty platform and a full one.
  6. Static view. Cross-side values change as the platform matures. Subsidies that make sense at launch may be wrong at scale, and vice versa; revisit the price structure.
  7. No regulation or fairness constraints. Below-cost pricing to one side can raise competition-law questions for dominant firms. The model does not answer them; counsel does.

How managers use it in practice

Map the sides and the interactions. List every group that joins, what each one wants from the others, and which interaction creates value (a sale, a booking, a robot mission). Sketch who would pay if the platform disappeared. This avoids the common error of treating complementors as suppliers.

Estimate cross-side values. Use pilots, A/B tests on fees, seller surveys ("how much more would you pay per year if buyer traffic doubled?"), cohort data (retention by network size) or conjoint analysis (mba/mba-conjoint-analysis-card). Even rough estimates of which side cares more about the other are enough to choose the subsidy side.

Set the price structure, then the level. Decide which side is the subsidy side (high cross-side value created, high price sensitivity, low cost to serve) and which is the money side (high value received, few alternatives). Then set levels with the formulas or a Solver model, and test them against rival platforms' fees.

Plan the launch path. Choose a micro-market, decide which side to seed, budget the subsidy until critical mass, and define the signal that critical mass has been reached (repeat interactions, liquidity: the share of listings that sell within a set time, or of searches that end in a match).

Choose the degree of openness and the governance. Decide which interfaces third parties may use, what they pay, how quality is certified and which data they see. Write the rules down: complementors invest only if they trust that the owner will not expropriate them.

Watch for multi-homing and envelopment. Track the share of each side that also uses rival platforms, and list the large platforms that share your users and could bundle your function. Lower multi-homing costs on your subsidy side make your subsidy less effective; exclusivity deals can raise them but carry legal risk.

Worked example A (fully solved): pricing a two-sided resale marketplace

Situation (fictional). Northfield Outdoor Co., the outdoor-gear retailer, plans a resale marketplace where customers sell used gear (tents, packs, skis) to other customers. Northfield handles payment, inspection labels and shipping labels. The team estimated annual participation demand on each side from a pilot in two regions (Example values, USD per member per year, members in thousands):

ParameterBuyers (B)Sellers (S)
Standalone value vv (top member, empty other side)20.0030.00
Cross-side value α\alpha0.10 per thousand sellers0.40 per thousand buyers
Heterogeneity tt0.200.50
Cost to serve cc4.006.00

Sellers care much more about buyers (αS=0.40\alpha_S = 0.40) than buyers care about sellers (αB=0.10\alpha_B = 0.10): a seller's main problem is finding someone who wants a used tent. Stability holds (tBtS=0.10>αBαS=0.04t_B t_S = 0.10 > \alpha_B \alpha_S = 0.04) and so does concavity (4tBtS=0.40>(αB+αS)2=0.254 t_B t_S = 0.40 > (\alpha_B + \alpha_S)^2 = 0.25).

Step 1: the profit-maximising prices. Maximising Π\Pi over pBp_B and pSp_S (Solver, or the first-order conditions) gives:

ResultValueExcel (B2:B5 buyer v, alpha, t, c; C2:C5 seller v, alpha, t, c; F2 pB, F3 pS, F4 nB, F5 nS, F6 profit)
Buyer price pB∗p_B^{*}USD -5.60Solver: maximise F6, changing F2:F3, unconstrained
Seller price pS∗p_S^{*}USD 46.00Solver result
Buyers nB∗n_B^{*}186.67 thousand=(C4*(B2-F2)+B3*(C2-F3))/(B4*C4-B3*C3)
Sellers nS∗n_S^{*}117.33 thousand=(B4*(C2-F3)+C3*(B2-F2))/(B4*C4-B3*C3)
Profit Π∗\Pi^{*}USD 2,901.33 thousand a year=(F2-B5)*F4+(F3-C5)*F5

Step 2: read the price formula. Check the buyer price with the three-term formula:

pB∗=cB−αSnS∗+tBnB∗=4.00−0.40×117.33+0.20×186.67=4.00−46.93+37.33=−5.60p_B^{*} = c_B - \alpha_S n_S^{*} + t_B n_B^{*} = 4.00 - 0.40 \times 117.33 + 0.20 \times 186.67 = 4.00 - 46.93 + 37.33 = -5.60

Each extra buyer is worth USD 46.93 a year to the seller side in total (USD 0.40 per thousand buyers to each of 117.33 thousand sellers, that is USD 0.0004 per seller), more than the buyer's cost plus markup (4.00 + 37.33 = 41.33). The negative price means Northfield pays each buyer about USD 5.60 a year net, on top of the USD 4.00 it spends serving them: a total subsidy of USD 9.60 per buyer relative to cost. In practice this is free membership plus benefits the buyer values at about USD 5.60 a year (a welcome credit, free return shipping), funded by seller fees. The seller price is 6.00−0.10×186.67+0.50×117.33=6.00−18.67+58.67=46.006.00 - 0.10 \times 186.67 + 0.50 \times 117.33 = 6.00 - 18.67 + 58.67 = 46.00: sellers pay because buyers matter to them and because sellers bring buyers relatively little.

Step 3: compare with pricing each side on its own. A team that prices each side as a separate product (pi=ci+tinip_i = c_i + t_i n_i, ignoring that its own price moves the other side) ends at buyer price USD 14.22 and seller price USD 28.22, with 51.11 thousand buyers, 44.44 thousand sellers and profit USD 1,510.12 thousand. Internalising the cross-side effects raises profit by USD 1,391.21 thousand (92.1%). Without any network effects (α=0\alpha = 0) the best possible profit would be USD 608.00 thousand (buyer price 12.00, seller price 18.00): the network effects are where most of the value sits.

Pricing approachpBp_BpSp_SnBn_BnSn_SProfit (USD k a year)
Two-sided optimum (Example)-5.6046.00186.67117.332,901.33
Each side priced separately (Example)14.2228.2251.1144.441,510.12
No network effects benchmark (Example)12.0018.0040.0024.00608.00

Step 4: what if a negative price is not practical? Paying members in cash invites fraud (fake buyer accounts). If Northfield sets pB=0p_B = 0 (free, no credit) and re-optimises, the best seller price is USD 39.00, with 151.67 thousand buyers, 103.33 thousand sellers and profit USD 2,803.33 thousand: only USD 98.00 thousand (3.4%) below the unconstrained optimum. The chart shows how flat the profit curve is near the optimum and how steeply it falls when buyers are charged.

Interpretation and recommendation. Make buyers the subsidy side: free membership plus a modest first-purchase credit, funded by seller fees (in practice a listing fee or a take rate equivalent to about USD 46 a year for an average seller). Prefer non-cash subsidies that only real buyers use. Re-estimate αS\alpha_S after a year: if sellers start to compete with each other for buyers, the seller-side value of more sellers turns negative and the structure shifts.

Worked example B (half solved): critical mass for a member community

Situation (fictional). Northfield also plans a paid trail-conditions community where members post and read real-time reports on trails, snow and river levels. Most members both read and post, so treat it as one side with a direct network effect. Market research suggests N=120N = 120 thousand potential members, standalone value uniform from 0 to V=12V = 12 USD a year (the map and offline guides are useful alone), and network value α=0.25\alpha = 0.25 USD a year per thousand members. A member joins if standalone value plus αn\alpha n is at least the price.

Step 1: equilibria at a price of USD 18 a year. Because p=18>V=12p = 18 > V = 12, nobody joins an empty community: zero members is an equilibrium. Full membership is also an equilibrium, because even the member with zero standalone value gets 0.25×120=30≥180.25 \times 120 = 30 \ge 18. Between them sits the critical mass:

n∗=120×18−12120×0.25−12=120×618=40 thousand members (33.3% of N)n^{*} = 120 \times \frac{18 - 12}{120 \times 0.25 - 12} = 120 \times \frac{6}{18} = 40 \text{ thousand members (33.3\% of } N)

Excel: =N*(p-V)/(N*alpha-V) with named cells. Above 40 thousand members, each new member raises the value for others enough to pull in more; below it, the community shrinks towards zero.

Step 2: other prices (complete the table).

Price pp (USD a year)Critical mass n∗n^{*} (thousand)Share of NNFull membership stable? (αN≥p\alpha N \ge p)
15 (Example)[ ][ ][ ]
18 (Example)40.0033.3%Yes (30≥1830 \ge 18)
24 (Example)[ ][ ][ ]
31 (Example)Not definedNot defined[ ]

Step 3: the cost of seeding. Northfield can give a free first year to enough loyalty-programme customers to pass critical mass. Cost to serve is USD 3 per member per year.

Price ppSeeding cost (free year for n∗n^{*} members)Annual contribution at full membership, N(p−3)N(p - 3)
15[ ]USD 1,440 thousand
18USD 120.0 thousandUSD 1,800 thousand
24[ ][ ]

Step 4: interpretation. [Explain why a higher price raises both the prize and the hurdle, and which price you would choose if Northfield can reliably seed only about 50 thousand loyalty members.]

Exercise C (unsolved): opening Corvane's fleet software

Situation (fictional). Corvane Robotics is considering opening its fleet-orchestration software to third-party robot makers (the platform idea in the mba/mba-strategy-guide mini-case, developed further in mba/mba-technology-strategy-guide). Two sides would join: warehouse sites (side W), which pay for orchestration, and robot makers (side M), which pay for certification and access to the interface. The team's stylised estimates (Example, USD thousand per member per year, members counted in units):

ParameterSites (W)Robot makers (M)
Standalone value vv20.010.0
Cross-side value α\alpha0.30 per robot maker0.60 per site
Heterogeneity tt0.500.60
Cost to serve cc4.03.0
  1. Check the stability and concavity conditions.
  2. Compute participation on both sides at prices of 10.0 for each side. Show the Excel formulas.
  3. Use Solver to find the profit-maximising prices, participation and profit.
  4. Which side is the subsidy side? Is its price below cost? Explain with the three-term formula.
  5. Sites may multi-home (run a rival's orchestration in another building). How would rising multi-homing by sites change your answer, qualitatively?
  6. Competitor A, the segment leader, could bundle orchestration free with its robots (envelopment). What would you recommend Corvane do in response?

Connections to other cards

  • mba/mba-strategy-guide: the core introduction to network effects and the Corvane platform idea.
  • mba/mba-technology-strategy-guide: the elective this card belongs to; its mini-case asks whether Corvane should open its platform, with NPV exhibits.
  • mba/mba-competitive-advantage-card and mba/mba-resource-based-view-card: a network is a resource; test it with VRIO, including whether rivals can reach the same users by multi-homing.
  • mba/mba-game-theory-card: platform competition as a game; commitment and expectations decide which equilibrium the market reaches.
  • mba/mba-market-structure-card: concentration after tipping; why platform markets can be both concentrated and contestable.
  • mba/mba-pricing-card and mba/mba-demand-elasticity-card: one-sided pricing, the building block of each side.
  • mba/mba-customer-lifetime-value-card: acquisition subsidies are justified by lifetime value plus the cross-side value a member creates.
  • mba/mba-innovation-portfolio-card: a platform bet is usually an adjacent or transformational project; fund it in stages.
  • mba/mba-new-product-diffusion-card: adoption curves for the launch phase.
  • mba/mba-ai-strategy-card: data network effects and when data creates advantage.

Common mistakes and misconceptions

  1. Calling every marketplace a platform. A reseller that sets prices and holds inventory is a merchant; platform logic applies only where the sides interact directly and choose to join.
  2. Pricing each side separately. Treating the sides as two products ignores how one side's price moves the other side's participation; in Worked example A this leaves 48% of the optimal profit on the table.
  3. Assuming the subsidy side is always the consumer. The subsidy side is the one that creates more value for the other side and is more price-sensitive. In some markets that is the developer or the seller.
  4. "Network effects mean winner takes all." Multi-homing, differentiation, congestion and local network effects (value depends on users near you, not on all users) often keep several platforms alive.
  5. Counting users instead of interactions. Investors and managers should track liquidity and repeat interactions, not registered accounts.
  6. Subsidising forever. Launch subsidies must have an end point linked to a measure of critical mass, or they become a permanent cost.
  7. Ignoring negative same-side effects. More sellers mean more competition among sellers; too many low-quality complementors drive users away. Curation is part of the product.
  8. Opening without governance. Opening an interface without rules on quality, data and pricing invites free-riding and low-quality complements; closing it later damages trust.
  9. Confusing scale economies with network effects. Falling unit cost with volume is a supply-side effect; network effects raise willingness to pay. Both can be present; they have different strategic consequences.
  10. Forgetting envelopment. The most dangerous rival may be a platform in an adjacent market that already has your users and can bundle your function.
  11. Treating below-cost pricing as legally neutral. For a dominant platform, subsidies, self-preferencing and exclusivity can raise competition-law questions; check with counsel.

Self-test questions

  1. Define a multi-sided platform and give one example of a business that looks like a platform but is not one.
  2. Distinguish direct from indirect network effects, with one example of each and one example of a negative effect.
  3. What did Rochet and Tirole (2003) mean by the importance of price structure?
  4. State the three terms of the monopoly platform pricing formula and explain each in words.
  5. A side has cost to serve 2, the other side has 40 thousand members who each value one more thousand members of this side at 0.5, and this side's markup term is tn=0.1×60t n = 0.1 \times 60. What is the optimal price?
  6. In the critical-mass model, N=50N = 50 thousand, V=10V = 10, α=0.4\alpha = 0.4 and p=14p = 14. Compute the critical mass and the highest price at which full membership is still an equilibrium.
  7. In Worked example A, how many buyers and sellers join at pB=0p_B = 0 and pS=39p_S = 39? Give the Excel formula for buyers.
  8. Name four launch tactics for the chicken-and-egg problem.
  9. What is the difference between granting access and devolving control (Boudreau 2010)?
  10. What is platform envelopment, and why can it defeat a rival with a better product?
  11. Give three reasons why a platform market may not tip to a single winner.

My notes and next steps

ItemMy answer
The platform I am analysing and its sides[Platform], [sides]
Which side I would subsidise and why[Side], [reason]
Evidence I need on cross-side values[Pilot, survey, data]
Launch tactic and critical-mass signal[Tactic], [signal]
Biggest multi-homing or envelopment threat[Threat]
Review dates[Two days], [one week]

Sources and further reading

Primary works.

  • Katz, M. L., and Shapiro, C. (1985). Network externalities, competition, and compatibility. American Economic Review, 75(3), 424-440.
  • Katz, M. L., and Shapiro, C. (1994). Systems competition and network effects. Journal of Economic Perspectives, 8(2), 93-115.
  • Arthur, W. B. (1989). Competing technologies, increasing returns, and lock-in by historical events. The Economic Journal, 99(394), 116-131.
  • Rochet, J.-C., and Tirole, J. (2003). Platform competition in two-sided markets. Journal of the European Economic Association, 1(4), 990-1029.
  • Caillaud, B., and Jullien, B. (2003). Chicken and egg: Competition among intermediation service providers. The RAND Journal of Economics, 34(2), 309-328.
  • Parker, G. G., and Van Alstyne, M. W. (2005). Two-sided network effects: A theory of information product design. Management Science, 51(10), 1494-1504.
  • Armstrong, M. (2006). Competition in two-sided markets. The RAND Journal of Economics, 37(3), 668-691.
  • Boudreau, K. (2010). Open platform strategies and innovation: Granting access vs. devolving control. Management Science, 56(10), 1849-1872.
  • Eisenmann, T., Parker, G., and Van Alstyne, M. (2011). Platform envelopment. Strategic Management Journal, 32(12), 1270-1285.
  • Zhu, F., and Iansiti, M. (2012). Entry into platform-based markets. Strategic Management Journal, 33(1), 88-106.
  • Cennamo, C., and Santaló, J. (2013). Platform competition: Strategic trade-offs in platform markets. Strategic Management Journal, 34(11), 1331-1350.
  • Hagiu, A., and Wright, J. (2015). Multi-sided platforms. International Journal of Industrial Organization, 43, 162-174.
  • Regulation (EU) 2022/1925 of the European Parliament and of the Council (Digital Markets Act), applicable from 2 May 2023; see the Commission's DMA site.

Books and textbooks (further reading only). Parker, Van Alstyne and Choudary, Platform Revolution (W. W. Norton, 2016); Shapiro and Varian, Information Rules (Harvard Business School Press, 1999); Evans and Schmalensee, Matchmakers: The New Economics of Multisided Platforms (Harvard Business Review Press, 2016); Cusumano, Gawer and Yoffie, The Business of Platforms (Harper Business, 2019); Schilling, Strategic Management of Technological Innovation (McGraw-Hill).

Open resource. OpenStax, Principles of Management (CC BY-NC-SA 4.0), chapters on strategy, for background.

Answer key

Worked example B.

Price ppCritical mass n∗n^{*}Share of NNFull membership stable?
15120×3/18=20.00120 \times 3/18 = 20.00 thousand16.7%Yes (30≥1530 \ge 15)
24120×12/18=80.00120 \times 12/18 = 80.00 thousand66.7%Yes (30≥2430 \ge 24)
31Not definedNot definedNo: 30<3130 < 31, even a full community does not hold members with low standalone value

Seeding cost: at p=15p = 15, 20.00×3=20.00 \times 3 = USD 60.0 thousand; at p=24p = 24, 80.00×3=80.00 \times 3 = USD 240.0 thousand, with annual contribution at full membership 120×(24−3)=120 \times (24 - 3) = USD 2,520 thousand. A higher price raises the prize (contribution at full membership) but also the critical mass, because each member's network value must cover more of the price. With about 50 thousand reliable seeds, USD 24 is out of reach (critical mass 80 thousand); USD 18 (critical mass 40 thousand) leaves a margin of 10 thousand members for seeds who do not engage. The middle equilibrium is unstable because the adoption response is steeper than one for one: each extra member draws in Nα/V=30/12=2.5N\alpha / V = 30/12 = 2.5 further members.

Self-test.

  1. A platform enables direct interactions between distinct groups that each choose to join it and keep control over the terms of their interaction (Hagiu and Wright 2015). A retailer that buys inventory and resells it, setting the prices itself, is a merchant, not a platform.
  2. Direct: a messaging network is more valuable the more people use it. Indirect: more sellers attract buyers and more buyers attract sellers. Negative: more sellers in one category compete for the same buyers.
  3. On a two-sided platform the volume of interactions depends on how the total price is split between the sides, not only on the total, so the platform must choose the structure deliberately.
  4. Cost to serve the member; minus the external benefit the member brings to the other side; plus a markup that depends on this side's price sensitivity.
  5. p=2−0.5×40+0.1×60=2−20+6=−12p = 2 - 0.5 \times 40 + 0.1 \times 60 = 2 - 20 + 6 = -12: a subsidy of 12 per member.
  6. n∗=50×(14−10)/(50×0.4−10)=50×4/10=20n^{*} = 50 \times (14 - 10)/(50 \times 0.4 - 10) = 50 \times 4/10 = 20 thousand. Full membership holds while p≤αN=20p \le \alpha N = 20.
  7. 151.67 thousand buyers and 103.33 thousand sellers. Buyers: =(C4*(B2-0)+B3*(C2-39))/(B4*C4-B3*C3), which is (0.5×20+0.1×(30−39))/0.06=9.1/0.06=151.67(0.5 \times 20 + 0.1 \times (30 - 39)) / 0.06 = 9.1/0.06 = 151.67.
  8. Any four of: seeding one side yourself, piggybacking on an existing user base, a single-side value proposition (a useful standalone tool), micro-market launch, marquee participants, subsidising the side that creates more value for the other.
  9. Granting access lets third parties build complements on the platform; devolving control gives up control over the platform itself (its rules or core technology). In Boudreau's data, granting access had the larger effect on the rate of new device development.
  10. Entry into a platform market by bundling the target's function into a multi-platform offer that shares users with it. The enveloper reaches the users at little extra cost and can price the bundled function low, so the rival's better product may not win enough users.
  11. Multi-homing at low cost, differentiation between platforms, congestion or negative same-side effects, local network effects, and regulation.

Revision history

VersionDateAuthorChange summary
0.1YYYY-MM-DD[Your name]First copy for [course, case or project]
0.2YYYY-MM-DD[Your name]Example: Replaced Northfield parameters with pilot estimates for our marketplace

Standard and licence note

Structure aligned with the economics and strategy literature on multi-sided platforms named above (Katz and Shapiro, Rochet and Tirole, Caillaud and Jullien, Parker and Van Alstyne, Armstrong, Boudreau, Eisenmann, Parker and Van Alstyne, Hagiu and Wright and others). The explanations, notation, examples, exercises and answers are original; no textbook, article or published case text is copied or paraphrased, and Northfield Outdoor Co., Corvane Robotics and every competitor named are fictional. This template is independent and not affiliated with or endorsed by any school, author, publisher or regulator. Regulatory facts are dated; check the current official text before relying on them. Template licence: MIT.

Modèles associés