Case Studies: eBay, Uber, Airbnb, Etsy, DoorDash | Marketplace Dynamics Tutorial

This chapter walks through five marketplaces that shaped the playbook: what worked, what broke, and what each one taught everyone who followed.

eBay: The Original Online Marketplace

Founded 1995. The template for almost everything that came after.

What worked

Auction format solved price discovery for unique, used, and rare items. A 1980 baseball card has no fixed price; an auction finds one
Feedback / rating system, early and bilateral. Buyers and sellers rated each other, and the platform displayed cumulative scores prominently. Trust infrastructure before most of the internet had any
PayPal acquisition (2002) locked in payments and dispute handling, fixing the trust gap in person-to-person transactions
Long-tail supply: eBay let millions of sellers list anything, creating a breadth no traditional retailer could match

What broke

Buy It Now and fixed-price listings eroded the auction's uniqueness and moved eBay closer to Amazon territory, where it competes poorly
Search quality has lagged. Matching was always eBay's weakness: the platform never made search feel modern, so buyers with clear intent went to Amazon
Fraud and counterfeits in some categories (luxury goods, electronics) damaged buyer confidence over years
Decline in certain verticals (cars went to dedicated platforms; collectibles to specialists)

The lesson

eBay showed that a marketplace can exist for tens of millions of unique items if you solve trust (ratings) and payments (escrow-like flows). It also showed that becoming the default for a category (fixed-price goods) is not the same as owning that category. Someone with better matching and logistics (Amazon) can dominate it.

Uber: Real-Time Matching and Subsidy Warfare

Founded 2009. Redefined how a hyperlocal real-time marketplace could be built.

What worked

Instant matching with GPS-based dispatch. The experience was "press a button, a car comes". Radical compared to calling a dispatcher and waiting
Dynamic pricing (surge) balanced supply and demand in real time, making the marketplace reliable even during spikes. Economically clean; brand-damaging
Operational launch playbook: a "general manager" per city, a fixed set of launch tactics (driver subsidies, rider promos, press), and metrics-driven scaling
Global expansion city by city, with playbook adaptation. Launched hundreds of cities in under a decade

What broke

Regulatory warfare. Uber fought taxi regulation in most cities. Won many, lost some. The reputational cost (Travis Kalanick's Uber era) left lasting scars
Unit economics took years to come to terms with. Drivers are classified as independent contractors in the US; rulings elsewhere (UK, EU) have been messier, changing the cost base
Eats diversification was a hedge against ride-share cyclicality; worked because Uber already had a logistics platform in place
Lost markets: Uber exited Southeast Asia, China, and Russia, folding into local players for equity stakes. Cost a lot, but clarified focus

The lesson

Uber proved the power of real-time, hyperlocal marketplaces and the brutal efficiency of the subsidy-led cold start. It also showed that aggression has long-tail costs. The playbook worked; the brand paid for it. Lyft's survival in the US is partly because Uber's early brand was uncomfortable for a large minority of users.

Airbnb: Trust as a Product

Founded 2008. Turned "would you let strangers stay in your home?" into a normal question.

What worked

Photography as a service. Early Airbnb sent professional photographers to host homes. A small operational cost produced listings that dramatically out-converted amateur ones. This single decision moved liquidity measurably
Trust infrastructure. Verified IDs, reviews (bilateral, simultaneous-reveal), and later AirCover insurance made the ask reasonable. Nothing about lodging stranger-in-your-home is naturally safe; Airbnb engineered the sense that it was
Brand. Airbnb spent heavily on brand, not just performance marketing. "Belong anywhere" positioned the platform as more than a cheaper hotel
Global demand from day one. Travel is inherently global; users look for lodging in cities they don't live in. This made Airbnb less local than Uber

What broke

Local regulation. Short-term rental bans or caps in many cities (New York, Barcelona, parts of Berlin, Paris). Airbnb has had to reshape supply repeatedly in response
Quality variance. As supply scaled, quality became uneven. "Aircover" and other trust investments were partly a response
Airbnb Experiences underperformed expectations. Lodging demand did not cleanly translate to experience demand
Hotel vs home tension. As Airbnb moved toward more professional hosts, some of the original charm (individual hosts) eroded, inviting regulatory pushback

The lesson

Airbnb showed that trust is not a policy; it is a product you invest in. It also showed that global demand with local supply creates a regulatory surface that never fully closes. A global marketplace in a regulated category is a perpetual negotiation.

Etsy: Long-Tail Supply in a Category Where Uniqueness Matters

Founded 2005. The marketplace for handmade, vintage, and craft goods.

What worked

Focus on a clear category (handmade/vintage) gave Etsy a specific brand and a supply side that didn't want to be on Amazon
Community tools. Etsy invested in seller forums, education, and events. Sellers stayed because the platform felt like home
Low-effort supply onboarding. Any maker could list. The variety made the platform discoverable for buyers looking for unique items
Long-tail matching. Search worked well enough that buyers with specific tastes kept coming back

What broke

Authenticity drift. Etsy relaxed its "handmade" definition in 2013 to allow manufacturers to sell on the platform. Community revolt. Eventually it clawed back some authenticity, but the tension persists
Competition with Amazon Handmade and others. Etsy's scale gave it durability, but the category is not exclusively theirs
Seller concerns about fees and ads. Etsy's take rate (including promoted listings) has climbed over the years. Sellers regularly organise complaints, sometimes strikes

The lesson

Etsy proved that a narrow category marketplace with strong community can defend itself against Amazon. It also showed that a community-centric platform has hard limits on how aggressively you can monetise before the community pushes back. Every take-rate increase costs something in trust.

DoorDash: Three-Sided, Hyperlocal, Logistics-Heavy

Founded 2013. A three-sided marketplace (customers, restaurants, couriers) with physical delivery as the core.

What worked

Dense suburban coverage where competitors (Grubhub, Uber Eats) were weaker. DoorDash won middle America before it won coastal cities
Tight operational control over the full loop: courier dispatch, restaurant integration, customer experience. More ops-heavy than a pure matching marketplace
Partnerships with major restaurant chains, sometimes exclusively. This locked in supply the competition couldn't easily dislodge
Aggressive scaling that burned cash but captured market share at a critical moment

What broke

Take rates on restaurants climbed past 30% in some cities, drawing regulation (California AB 286, NYC cap at 15%). Restaurants, especially small ones, publicly complained that DoorDash made them unprofitable
Courier classification (contractor vs employee) is an ongoing legal question. Changes would raise costs significantly
Profitability has been hard-won. The business works at high volume but remains thin on margin, heavily dependent on advertising revenue from restaurants buying visibility
Commoditisation risk: food delivery feels similar across platforms; switching costs are low

The lesson

DoorDash illustrated that marketplaces can be brutal operational businesses when physical logistics are part of the loop. It also showed that a three-sided marketplace has three groups to keep happy, and you can lose trust from any one of them. The restaurant backlash of 2020 onward is a case study in how quickly supply-side dissatisfaction can damage a brand.

Cross-Cutting Observations

Five marketplaces, a few shared themes:

All five invested heavily in trust before the market demanded it. eBay's feedback system, Airbnb's photography and insurance, Uber's rating and insurance, Etsy's verification, DoorDash's dispute flows
All five had painful cold-starts solved by some form of focused, non-scalable work. eBay manually curated early listings. Airbnb co-founders travelled to photograph hosts. Uber's early drivers were recruited by founders' personal networks
All five had take-rate tensions. Raising rates always hurt supply somewhere. Managing this without collapse is a perennial challenge
All five faced regulatory pressure in at least one major market and had to adapt
All five grew by concentrating on density before reach, even when it felt slow

Marketplaces That Didn't Make It

Briefly, for balance:

Homejoy (home cleaning marketplace). Folded 2015. Low retention, worker classification lawsuits, over-reliance on discounts
Sidecar (ride-sharing). Lost the war to Uber and Lyft, who had better capital and more aggressive expansion
Zirtual (virtual-assistant marketplace). Unit economics didn't work; abrupt shutdown in 2015
Munchery (food marketplace, vertical). Kitchen-to-consumer model never reached liquidity before running out of runway

These aren't failures because they were dumb ideas; they're cautionary because the execution and economics didn't hold. Every one had moments of real promise.

The Patterns to Steal

Pick a narrow, high-density seed market
Overinvest in trust early
Subsidise the constrained side, but budget it carefully
Do whatever doesn't scale until the network does
Measure liquidity, not vanity metrics
Expand city by city, not everywhere at once
Take your time raising take rates
Don't believe your own press

Next Steps

Continue to 12-best-practices.md for the patterns and anti-patterns that emerge across every marketplace story.