Marketplaces & AI

Not Gonna Make It

A lot of software and online business models are in deep trouble with the advent of AI.

• Enterprise Software / SaaS: The lower cost of custom coding provided by LLMs is helping companies bring key systems in-house. They can build and maintain bespoke software at a lower cost.
• Information Sites / Ad-Supported Content: Search engines are now summarizing and presenting information directly, removing the need to click on ad-supported content sites.
• Review Sites: AI models can generate hyper-realistic, yet fake, reviews on demand, destroying trust in platforms reliant on user-generated feedback.
• Content Generation: While AI is a tool, it directly competes with human content creators, pushing down the price and demand for generic human-written content.
• Social Media: Fake accounts and content has already been an intractable problem for social media companies, and AI accelerates the scale of the problem.
• Software Development Agencies: AI-powered coding assistants and LLMs accelerate development so much that the need for large, external development teams is shrinking rapidly.

Survive & Thrive

What online businesses have a bright future? Marketplaces.

• Specialized Job Boards: Connect specific talent to niche roles, valuable amidst industry shifts. (Example: Nanny Lane)
• Services Matching: Facilitate local supply and demand for essential personal services. (Example: Housekeeping services)
• Dating Platforms: Match individuals for personal relationships, a task AI cannot replicate. (Example: Hinge)
• Unique E-commerce: Curate and connect buyers and sellers of niche, handmade, or vintage items. (Example: Etsy)
• Investment/Fundraising Marketplaces: Match capital (supply) with vetted, growing companies (demand). (Example: AngelList)
• Tutoring/Education Marketplaces: Provide personalized learning and expert human guidance. (Example: Outschool)
• Real Estate Brokerage/Listings: Facilitate high-stakes property transactions requiring local knowledge and trust. (Example: Zillow)
• Physical Goods Rental: Match local, inspected assets (e.g., equipment) with temporary demand. (Example: Rent almost anything platform)
• Event Ticketing (Secondary/Resale): Securely manage transactions for high-demand, finite inventory with authentication. (Example: StubHub)
• Car Sharing/Peer-to-Peer Rental: Match vehicle owners with local short-term renters with built-in trust and logistics. (Example: Turo)
• High-Skilled Freelancer Marketplaces: Vet and manage secure transactions for highly specialized professional expertise. (Example: Upwork for legal/engineering)

This is not to say that AI won’t dramatically change the operations of marketplace companies, but those apply to all companies. AI systems are replacing customer support and software development roles first, with lots more to come. But crucially it is not a threat to the business models of marketplace companies in the same way that it is to others.

User Acquisition

The biggest effect of AI on user acquisition for marketplace companies so far is the shift of user behaviour from Search (Google, Bing) to Chat (ChatGPT, Gemini, Claude). SEO is not dead, but it is no longer the future.

Onboarding

AI can make onboarding easier, by inferring more. Posting a classified ad these days has never been easier as modern marketplaces will write your ad for you, often based on a single photo.

Matching

Believe it or not, major matching platforms (like Match Group’s Plenty of Fish) have little to gain from LLMs. They have been pioneering extremely sophisticated Machine Learning algorithms for a decade, that are today highly honed and unbeatable by LLM-based approaches.

In job marketplaces, geospatial proximity, recency, and faceted search are the pillars. Job seekers needs to see listings that are near them, newly posted, and with job titles they are targeting. Modern AI-enabled platforms like Cerulean do use LLMs as a component of matching, but the other pillars remain.

Monetization

Marketplace businesses make money in many different ways, from ads, to premium profiles/listings, to payment intermediation. After matching families with nannies on CanadianNanny.ca, for example, CareGuide funnels new matches to HeartPayroll, which processes $25M/year in caregiver payroll. So far, revolutions in AI has shown little effect on any of these monetization methods.