Archetype 5: 2-Sided Marketplace

By Ryan Vanshur

Also published on the Guild Letter: Read the original issue.

Subscribe now If you’re building a two-sided marketplace and wondering why nobody’s using it despite your beautiful UI and compelling value prop, you’ve discovered the chicken-and-egg problem that kills 80% of marketplace startups. Buyers won’t join without sellers. Sellers won’t join without buyers. You need both simultaneously. And unlike normal SaaS where you can sell to one customer at a time, marketplaces are all-or-nothing. Either you achieve liquidity (enough supply that demand finds what they want) or you’re running an empty platform that nobody uses. This is Faire connecting wholesale brands with retailers. Mindbody linking fitness studios with consumers. Companies where network effects create winner-take-most dynamics, and the largest player gets exponentially more valuable while everyone else fights for scraps. Let me explain when marketplaces work, how to solve the cold-start problem, and why most vertical SaaS companies shouldn’t try this archetype. The Diagnostic: Should You Actually Build a Marketplace? Before you build a two-sided platform and burn $10M trying to reach liquidity (expensive mistake if the market doesn’t support it), let’s figure out if this is you. You should build Archetype 5 if: You have two distinct user types who must transact with each other. Not just “users and admins.” Fundamentally different personas. General contractors need subcontractors. Retailers need wholesale suppliers. Studios need consumers. If you only have one user type, you don’t have a marketplace. Coordination is genuinely hard without your platform. Finding the right subcontractor for a specialty trade on a complex construction project? Hard. Matching wholesale brands with boutique retailers who want their products? Hard. If users can easily find each other on Google or LinkedIn, they don’t need your marketplace. The market is fragmented on both sides. Thousands of suppliers, thousands of buyers. If there are only 50 suppliers and 100 buyers in your entire market, this isn’t marketplace-scale. That’s just a small industry with limited growth potential. Network effects are strong and defensible. More suppliers = more valuable for buyers. More buyers = more attractive for suppliers. This creates a flywheel. If adding users doesn’t make the platform meaningfully better for existing users, you don’t have network effects. Winner-take-most dynamics exist. Marketplaces tend toward consolidation. The largest player captures 60-80% of the market because network effects compound. If your market naturally supports 5+ equally-sized competitors, maybe it’s not a true marketplace. If 4 out of 5 are true, you might be able to build Archetype 5. If fewer than 3 are true, build a different archetype and stop pretending you’re a marketplace. Why Marketplaces Are Brutal to Build Here’s what nobody tells you about marketplaces: They take 3-5x longer to reach profitability than normal SaaS. Normal SaaS: Sign up customers one at a time. Each customer generates revenue immediately. Compound linearly. Marketplaces: Need critical mass before anyone gets value. Burn money for 18-36 months reaching liquidity. Then suddenly hit inflection point and grow exponentially. The cold-start problem is real. Your first 100 suppliers join a platform with zero buyers. They list their products or services. Nobody buys. They leave. You’re back to zero. This is why 80% of marketplaces fail in the first 2 years. They never solve the cold-start problem. They run out of money before reaching liquidity. The 20% that survive? They become incredibly valuable because network effects create moats competitors can’t cross. The Three Strategies That Actually Solve Cold-Start You can’t solve the chicken-and-egg problem with marketing. You solve it with strategy. Here are the three that work. Strategy 1: Subsidize One Side Make one side free. Charge the other side. The free side grows fast. The paid side joins because the free side is there. Procore does this. Subcontractors use Procore for free. General contractors pay $200K-$1M annually. Why? Because all their subs are already on Procore. Switching costs are enormous. Faire does this. Retailers join free. Brands pay 15-25% commission on sales. Why do brands accept this? Because 800,000 retailers are on the platform they can’t reach otherwise. The Key: Choose which side to subsidize carefully. Usually subsidize the side that’s harder to monetize or easier to acquire at scale. Strategy 2: Build Single-Player Mode First Create functionality that works for one side even without the other side. Then layer in the marketplace. Procore started as project management software for general contractors. Contractors got value managing projects even without subs on the platform. Then Procore added the marketplace (inviting subs to projects). Contractors already loved the product, so they invited subs. Subs joined because contractors they work with asked them to. This solves the cold-start problem because you’re not trying to launch both sides simultaneously. Build for one side first. Prove value. Then add the second side. Strategy 3: Manual Curation of First 100 Users For the first 100 suppliers and first 100 buyers, recruit manually. Don’t try to scale. Just prove the model works. Airbnb ’s founders personally photographed listings in San Francisco. They didn’t build a self-serve photo upload tool. They did it manually to ensure quality. Uber recruited first 100 drivers in San Francisco by visiting them in person. Gave them iPhones. Trained them personally. Faire recruited first 100 boutique retailers in LA manually. Then recruited 50 brands that matched their aesthetic. Proved the marketplace worked. Then scaled. The Lesson: Don’t automate until you have proof. Manual curation ensures quality and lets you learn what actually creates successful matches. Subscribe now The Playbook: How to Actually Execute This Okay, you’ve decided to build a marketplace. Here’s how to not burn $10M on your way to failure. Step 1: Pick one city or one category. Total. Don’t launch nationwide. Don’t try to serve every industry segment. Pick one geography or one product category and dominate it before expanding. Uber launched in San Francisco only. Dominated SF. Then expanded city by city. Faire launched with home goods only. Dominated home goods. Then expanded to fashion, beauty, food. The Reason: Liquidity is geographic or categorical. You need enough density in one area that buyers find what they want. Spreading yourself thin means low liquidity everywhere = nobody gets value. Step 2: Define what “liquidity” means with a number. Liquidity isn’t a feeling. It’s a metric. How many suppliers do you need so that 80% of searches return 5+ relevant results? For Faire (wholesale marketplace): A retailer searching “handmade candles” should find 30+ relevant brands. Otherwise the marketplace feels empty. For Procore (construction): A GC should be able to find 5+ qualified subcontractors for any specialty trade in their region. Measure this religiously. You’re not at liquidity until this number hits your target. And you won’t grow exponentially until you hit liquidity. Step 3: Obsess over quality, not just quantity. 1,000 mediocre suppliers is worse than 100 great suppliers. Buyers try the marketplace, find low-quality matches, leave and never come back. How to Ensure Quality: Manual curation for first 500 suppliers (reject bad ones before they damage the experience) Ratings/reviews (let buyers flag low-quality suppliers) Quality thresholds (minimum standards for listings, photos, response time) Faire curates aggressively. Only 30-40% of brand applications get approved. This maintains quality and makes buyers trust the platform. Step 4: Build trust mechanisms from day one. Marketplaces are trust platforms. Buyers need to trust that suppliers won’t scam them. Suppliers need to trust they’ll get paid. Essential Trust Features: Verified badges (this supplier has been vetted) Ratings and reviews (transparent feedback) Payment protection (platform holds money in escrow until service delivered) Dispute resolution (platform mediates when things go wrong) Insurance/guarantees (platform covers losses from fraud) Without these, your marketplace is just Craigslist. And Craigslist is where you get scammed. Step 5: Use AI to accelerate liquidity by 60%. AI changes the marketplace cold-start timeline from 24 months to 9 months. Here’s how. AI-Powered Supply Recruitment: Scrape the web for potential suppliers. Score them automatically (follower count, engagement rate, product quality from photos). Rank top 1,000. Send personalized outreach. Recruit 10x faster than manual outreach. AI-Powered Matching: Instead of keyword search returning 500 random results, AI recommends top 10 suppliers based on buyer’s past behavior, preferences, location. Increases match quality 3-5x. AI-Powered Activation: New supplier joins. AI analyzes their profile. AI suggests: “Your photos are too dark. Want me to enhance them?” AI optimizes pricing based on similar successful suppliers. Activation rate increases from 40% to 75%. This isn’t theory. Faire and others are doing this in production right now. The Economics: Why This Is Winner-Take-Most Let’s talk about why successful marketplaces are worth $5-50 billion while failed marketplaces are worth zero. Marketplace Economics at Scale: Supply: 100,000 suppliers Demand: 800,000 buyers GMV (Gross Merchandise Volume): $3 billion annually Take rate: 20% (commission from suppliers) Revenue: $600 million Gross margin: 85%+ (mostly software, minimal fulfillment) CAC: Drops to near-zero once network effects kick in (both sides recruit each other) Retention: 95%+ (locked in by network effects) Compare to normal SaaS at $600M revenue: Need 60,000 customers at $10K ACV Each customer acquired individually (no viral growth) CAC stays constant or increases as you scale Retention: 90-95% (good but not exceptional) Marketplaces are harder to start but more valuable at scale because of compounding network effects. The Metrics That Actually Matter Your investor deck shows GMV growth. Cool. Here’s what actually predicts marketplace success: Liquidity score: What percentage of searches return 5+ relevant results? Target: 80%+. Below 60% and your marketplace feels empty. Repeat transaction rate: Are buyers coming back monthly? Target: 3-5 transactions per buyer per year. One-and-done buyers mean your matching is bad. Supply-side retention: What percentage of suppliers are active after 12 months? Target: 70%+. Below 50% means suppliers aren’t making money on your platform. Take rate sustainability: Can you maintain 15-25% take rate or are suppliers bypassing you? If suppliers are finding ways to transact off-platform, your take rate is too high or your value is too low. Cohort economics: Are newer cohorts better or worse than old ones? If newer cohorts have lower retention or lower GMV per user, you’re scaling too fast and quality is suffering. The Mistakes Everyone Makes Mistake 1: Launching both sides simultaneously. You try to recruit suppliers and buyers at the same time. Both sides see an empty platform. Both sides leave. Pick one side first. Mistake 2: Expanding geographically too fast. You launch in 10 cities. None of them have liquidity. All of them feel empty. Should’ve dominated 1 city first. Mistake 3: Optimizing for growth over quality. You approve every supplier to boost supply numbers. Quality drops. Buyers have bad experiences. They churn and tell others the platform sucks. You traded short-term growth for long-term reputation damage. Mistake 4: Underestimating time to liquidity. You budgeted 12 months and $3M to reach liquidity. It actually takes 24 months and $8M. You run out of money before the flywheel spins. Game over. The AI Multiplier (Because It’s 2026) AI transforms marketplaces in three ways: AI cuts cold-start time 60%. Automated supply recruitment, smart matching, activation assistance. What took 24 months now takes 9 months. AI increases match quality 3-5x. Better recommendations mean higher transaction rates, better retention, more GMV per user. AI detects fraud 10x better. Fake suppliers, low-quality sellers, scammers. AI flags them during onboarding before they damage the platform. Fraud rate drops from 5% to 0.5%. When NOT to Build a Marketplace Real talk: Most vertical SaaS companies shouldn’t build marketplaces. Don’t build Archetype 5 if: Only one user type → Not a marketplace, you’re normal SaaS Users can easily find each other without you → No coordination value Market supports 5+ equally-sized competitors → No network effects or winner-take-most dynamics You have <$10M capital → You’ll run out of money before reaching liquidity You’re impatient → This takes 3-5 years to work, not 18 months Marketplaces are high risk, high reward. When they work, they’re incredibly valuable. When they don’t, you burn $10M and have nothing to show for it. The Bottom Line Two-sided marketplaces create winner-take-most dynamics. The largest player gets exponentially more valuable while smaller players fight for scraps. If you can solve the cold-start problem and reach liquidity, you’ve built a business with network effects competitors can’t replicate. That’s worth $5-50 billion. If you can’t solve cold-start, you’ve built an empty platform nobody uses. That’s worth zero. The difference between success and failure is usually: Did you focus on one geography/category first? Did you subsidize the right side? Did you manually curate quality in the early days? The companies that nailed this (Procore, Faire, Airbnb, Uber) are worth billions. The companies that tried and failed (hundreds you’ve never heard of) burned investor money and quietly shut down. Pick your strategy carefully. Execute with discipline. And don’t expand until you’ve dominated your first market. Your call. Next Week: Archetype 6 — The Hybrid Growth Engine (Or: How to Build Three Companies at Once) SaaS + marketplace. B2B + B2C. Subscriptions + take rates. Multiple revenue engines working together…and one very confused marketing team. When it works, it’s the most defensible model in vertical SaaS. When it doesn’t, your CFO starts drinking at 11am. We’ll break down who actually pulls this off, why the operational complexity kills most teams who try it, and the honest diagnostic for whether you should even attempt it. If Archetype 5 was hard mode, Archetype 6 is building the whole arcade. Ryan