IGNITE EXTENDED / LESSON 11
When does growth strengthen a business model, and when does it not?
A network-effect map and an adoption-risk memo.
A company can become larger without making its service more valuable to an existing user. A network effect requires a specific connection: additional relevant participants or activity improve the experience for someone already using the system.
Direct effects arise within a participant group. Indirect effects connect groups, such as hosts and guests. The Airbnb material makes the distinction concrete. Additional listings can improve a guest’s choice, while additional relevant demand can make hosting more worthwhile. Poor matches, congestion and low-quality supply can weaken either effect.
A matching service may need useful participation in a particular place, category or time window. The total number of users across the company can conceal thin local markets. Participants in one city may provide little immediate value to someone searching in another.
The route and listing diagrams in the source illustrate this concentration. Expansion should be evaluated at the level where matching occurs. Building density in a narrow market can provide a more useful experience than spreading the same acquisition effort across many markets.
Scale concerns the amount of activity. A network effect concerns whether additional participants change the value experienced by other participants.
Unit cost can fall as output grows.
The value to a participant changes as other relevant participants join or use the system.
Illustrative application: buying more advertising can increase marketplace registrations without improving matching. Adding relevant hosts in a location may improve a guest’s choice, while poor-quality or irrelevant listings can add little. Examine the experience of existing participants rather than treating a larger count as the mechanism.
A defensible network claim names who joins, whose value changes and the interaction causing that change.
Indirect effects depend on relevance, quality, location and successful matching.
The adoption material distinguishes innovators, early adopters and later market groups. Early customers may tolerate inconvenience because they strongly value a new capability. Later customers may require reliability, compatibility, familiar support and clearer proof.
A successful early launch therefore does not establish that the next customer group will adopt for the same reason. Investigate the use situation and the complete solution the next group expects. Product changes, channel changes and a different explanation may all be necessary.
Registrations, active users, successful transactions and revenue measure different events. A platform can acquire many registrations while failing to create exchanges. Revenue can grow through higher prices even if participant activity stagnates.
Connect each metric to the mechanism being evaluated. For a matching service, examine relevant supply, search success, repeat use and participant economics. Historical Airbnb charts describe the period shown in the archive. Their value here is the reasoning about adoption and growth, rather than a claim about the company’s current performance.
Illustrative comparison: the same 12 participants can form a useful local cluster or four thin markets.
An early group can accept a product for reasons that do not transfer to the next group. Expansion therefore tests the adoption process as well as product appeal.
Identify the strongly felt problem that makes an incomplete solution worthwhile. A specialist may tolerate manual setup because the product solves a costly task that familiar alternatives handle poorly.
Examine the new buyer’s expectations and constraints. A mainstream team may require references, procurement approval, reliable support and compatibility with existing work before agreeing to try the same product.
Find the missing part of the complete offer. Better advertising cannot replace installation capacity or a required integration. Distinguish weak demand from a delivery obstacle.
Provide the missing support in a limited target group and measure activation and retained use. Track added costs as well as customer counts before assuming expansion improves the model.
Customer growth can come from acquisition spending, better retention or stronger interactions between users. The following tests separate these mechanisms.
Click either page to read it at full size. In the reader, use ← → to switch between these two pages.
Click either page to read it at full size. In the reader, use ← → to switch between these two pages.
Selected reading: PDF pages 2, 3, 4, 5
The selected chronology shows a marketplace developing through narrow launches, direct observation and changes to the transaction experience. Growth follows specific efforts to make hosting and booking work.
Event-driven launches create a concentrated setting in which accommodation supply and demand can meet. Early results vary across attempts.
The awkwardness of face-to-face payment prompts a change to payment handling. Matching people is only one part of enabling an exchange.
The chronology includes work on listing presentation and professional photography. The quality and relevance of available supply affect the guest’s experience.
The founders face limited cash and investor rejection while changing the offer and learning from customer contact. The case provides episodes to analyze, rather than a formula.
Selected reading: PDF pages 2, 3, 4, 5
More listings or bookings can follow publicity or hands-on work. A network-effect claim needs evidence that additional relevant participants improve value for others.
A retrospective chronology can make a path look inevitable. Identify the choices and uncertainty visible at each stage before using the outcome to judge them.
Marketplace liquidity depends on matching a particular need, place and time. A worldwide participant count can obscure weak availability in a target situation.
In a lodging marketplace, a guest needs a suitable place in the right location and time. Hosts need relevant demand and a workable exchange. A large global listing count does not guarantee a useful local match.
Where should an early network build density, and why?
In May 2025 Airbnb relaunched its app and added Services in 10 categories (chefs, massage, photography and others) in up to 260 cities and Experiences in 650 cities. The bet is that an existing guest base can attract a new supplier side. Whether that works will be decided city by city and category by category, which is this lesson's density question again.
EU Regulation 2024/1028 on short-term rentals applies from 20 May 2026. Platforms must collect and display host registration numbers, spot-check them against public registers, remove listings that authorities flag as invalid and share activity data with authorities. Supply quality and trust, which this lesson treats as part of network value, are increasingly a legal obligation rather than a product choice. Application depends on member-state implementation; check current rules.
Sources: Airbnb Newsroom: 2025 May Release, now you can Airbnb more than an Airbnb · TravelPulse: Airbnb launches new app, services and experiences features · European Commission Transition Pathways: EU regulation on data collection and sharing for short-term accommodation rentals · MCS: EU Regulation 2024/1028 on short-term rentals enters into application on 20 May 2026
Fictional service marketplace: the team can recruit 12 providers in one city or 3 providers in each of four cities. Customers need a nearby provider available on short notice. Quality and acquisition costs are otherwise assumed equal.
Compare the two patterns, specify the matching benefit you expect and identify evidence that could overturn the choice.
Relay ends FY1 with 60 clinics in Texas (illustrative). An investor calls it 'a two-sided marketplace with network effects' and asks why Relay does not race into ten cities. The second side is patients on waitlists. A typical clinic has about 42 empty slots a week. Relay's 40% fill rate assumes a waitlist deep enough to match patients to short-notice slots. Patients join only their own clinic's list; independent clinics will not share patients with competitors. Maya wants to open Houston next.
Produce a one-page network-effect map: who joins, whose value rises, at what level (clinic, group, city), what weakens the effect. Then write a Houston launch plan that solves the chicken-and-egg problem and names the metric that would overturn it.
'60 clinics' or '10,000 patients on waitlists' says nothing about value. For Relay the evidence is fill rate per clinic as waitlist depth rises, and net visits gained after moved-up patients' old slots are counted.
Airbnb's guests travel from anywhere to fixed supply, so part of its network is global. When both sides of a network stay in one clinic or city, scale elsewhere does not help, copycats can attack market by market, and the market does not tip to one winner.
Three pilot clinics run by enthusiastic managers tolerated Maya texting from a shared phone. A 40-location group needs a BAA, security review, EHR integration and references; more promotion will not substitute for those.
A Relay network-effect map (clinic-bounded waitlist effect, group-level pooling hypothesis, data effects kept separate) and a Houston launch plan with seeded waitlists and a fill-rate-versus-waitlist-depth metric.
Next lesson: Markets, information, incentives & causal evidence
Check your understanding
Question 1
Relay reaches 60 clinics across Texas. A pitch draft says: 'Every new clinic makes Relay more valuable for all clinics, so growth compounds into a winner-take-all position.' Which response is most accurate?
Patients join their own clinic's waitlist and rarely travel to a competitor, so added clinics do not raise each other's fill rates. Relay has a second side (patients), but its effect is local. Scale may cut costs or improve prediction models, which is a different mechanism.
Question 2
Illustrative pilot data: a clinic with 42 empty slots/week fills 20% with 10 active waitlisted patients and 35% with 30 patients, at $110 per visit. About how much weekly revenue do the extra 20 patients add?
Fills rise from 42 × 0.20 = 8.4 to 42 × 0.35 = 14.7, a gain of 6.3 visits. 6.3 × $110 ≈ $693/week. $1,617 is the total at 35%, not the increase. A per-clinic curve like this is the evidence for a local network effect.
Question 3
A food-delivery startup can spend its launch budget recruiting 12 restaurants in one neighborhood or 3 in each of four neighborhoods. Customers want delivery within 30 minutes. What is the strongest reason to choose the dense option?
Matching happens where delivery is feasible. Twelve nearby restaurants give real choice; three per area leave most requests poorly served. Reach is useless if each area is too thin to satisfy customers. The cost claim may or may not hold and is not the mechanism.
Question 4
Airbnb listings grew sharply in a city while bookings per listing fell. A board member concludes the network effect is strengthening. What should you check first?
A network effect requires added supply to raise value for existing participants. Falling bookings per listing suggests irrelevant or poorly located supply, or that demand did not keep pace. Registrations and price are different metrics and cannot show the mechanism.
Question 5
Relay's early pilot clinics loved the product despite manual setup. Mainstream multi-location groups now stall after demos. Which explanation best fits the adoption-gap reasoning in this lesson?
Early adopters tolerate gaps because the problem hurts them. Mainstream buyers want the complete solution and peer proof. A large value-to-price ratio does not close that gap, and more features do not replace missing integration or compliance work.
Question 6
Which observation would best support a genuine indirect network effect on a two-sided marketplace?
The test compares similar tasks under different participation levels and measures a benefit to existing users. Advertising-driven growth, price-driven revenue and funding are all growth without evidence that added participants improved anyone's experience.
Workbook
Module output: A business-model map, market/game models and an advantage-and-pivot memo.
Library
10 appear in the slides with notes; the rest are kept here for reference.
Source S20 · Mendelson · 10 pages
Source S19 · Mendelson · 14 pages
