The Death of Search: How Shopping Will Work In The Age of AI - YouTube

The Future of Search and Shopping in the Age of AI: E-commerce Evolution and Business Model Adaptation

The e-commerce landscape is undergoing a profound transformation with the rise of AI technologies, particularly Large Language Models (LLMs) and AI Agents. Traditional search paradigms are facing disruption, and both consumer behavior and business models are being reshaped. This article delves into how AI is challenging the existing order and foreshadowing the evolution of future shopping experiences.

Is the Current Internet Healthy? How Will AI Intervene?

The current state of the World Wide Web is precarious, riddled with a deluge of low-quality, SEO-driven "junk information." Alex points out that the declining quality of internet content means that "summarizing that crap is not helpful." This unhealthy situation manifests in two primary ways:

• The "Walled Garden" Effect: Much high-quality content is isolated within specific platforms, such as Twitter (now X) or Facebook, making it inaccessible to traditional search engines.
• Information Proliferation Driven by Excessive Commercialization: In the past, bloggers shared content out of passion; today, the emergence of numerous "Top 10" list websites often serves to generate Affiliate Revenue, relying on cheap labor to produce low-quality content optimized for SEO.

In this context, AI Agents are highly anticipated, with the potential to "declutter" this content by guiding people towards purchasing activities. Justine adds that while high-quality video content (e.g., YouTube) is currently difficult for traditional search engines to capture, in the future, LLMs could transcribe and analyze these videos to offer more precise recommendations to consumers, thereby improving the information environment.

How Will Google's Business Model Be Challenged by AI?

Google, as a search giant, faces severe challenges to its business model from AI. Alex likens Google to "a tax on GDP" because it extracts a percentage from vast consumer spending through its pay-per-click model. However, personal search queries, particularly in non-commercial domains, are increasingly being siphoned off to AI tools like ChatGPT.

In Google's "freemium" model, free searches (e.g., historical queries) are being diverted to ChatGPT because these queries lack commercial monetization potential. While paid searches (e.g., for a "tennis racket") currently remain a strength for Google, as paid ad results are often relevant to organic search results and enhance the search experience, Justine warns that early LLMs had "hallucination" issues in product recommendations (suggesting non-existent or inaccurately priced products), which temporarily drove users back to Google or Amazon. However, as companies like OpenAI work to improve AI's accuracy in e-commerce, Google faces an increased risk of losing more search queries.

What Role Will AI Play in Shopping Decisions?

AI will play different roles across various types of purchasing behaviors. Justine categorizes purchasing actions as:

• Impulse Buys: For instance, buying a Coca-Cola at the checkout counter. These purchases typically involve no prior research and are made instantaneously, making them difficult for AI to completely disrupt.
• Considered Purchases: Such as buying a house, wedding venue, or car, which are high-value items. Even if AI can assist with preliminary research, consumers usually still require in-person experience and interaction with human experts, making it challenging for AI to handle these end-to-end.

However, AI demonstrates immense potential in "intermediate-type purchases" that fall between these two extremes:

• Research Assistance: AI Agents can conduct in-depth research for busy consumers, such as finding the best travel handbag that fits a laptop and water bottle, synthesizing authentic user feedback from TikTok videos and Reddit posts.
• Price Optimization: AI Agents can continuously monitor the best prices for specific goods (e.g., laundry detergent) across the web and automatically purchase them when the price is right. For example, if an item is 30% cheaper on a specific website than elsewhere and can be delivered in a reasonable timeframe, AI could execute the purchase automatically.
• In-depth Personalized Consultation: For higher-value items like bicycles, sofas, or laptops, AI Agents can engage in dynamic, in-depth conversations with users, understanding all criteria, providing customized purchasing advice, and even acting like a human expert. One example involves teenage girls uploading street style photos of Lana Del Rey or Taylor Swift, and AI identifying the brand of hairpins or sweaters they are wearing, then offering more affordable alternatives.

Furthermore, the presence of a Universal Product Code (UPC) is a crucial dimension for gauging AI's potential in assisted purchasing. For products with a UPC (e.g., Nike shoes), AI can significantly optimize "lowest price" algorithms, automatically searching for the best deals, cashback, and coupons. This will automate manual processes, benefiting consumers who "value money more than time". For products without a UPC (e.g., a bar stool from Wayfair), the purchasing process is more complex, and while AI might assist in the initial research phase, direct automated purchasing remains challenging.

How Will Traditional E-commerce Attribution Challenges Evolve in the AI Era?

The e-commerce penetration rate has only reached 16% of total retail sales over the past 20 years, a figure that may underestimate the impact of online research on offline purchases. Justine points out that many consumers engage in "Research Online, Purchase Offline" (ROPO), conducting online research before completing their purchases in physical stores. For instance, when buying a laptop, users research on online platforms like Reddit, Instagram, and Apple's official website before visiting a physical store to experience the product firsthand.

Attribution is one of the most difficult problems in e-commerce, especially under the "Last-Click Attribution" model. Alex criticizes tools like Honey for "stealing attribution" by offering coupons just before a user makes a purchase, redirecting them and embedding a cookie, thus misleading e-commerce companies into believing these tools are the primary sales drivers. In the AI era, attribution will become even more complex, as consumers may interact with products across multiple touchpoints—Reddit, Super Bowl ads, ChatGPT—making it difficult to fully attribute a final purchase to any single AI agent. Amazon, however, is smarter in this regard, avoiding such attribution theft due to its powerful proprietary channels. Nevertheless, Amazon itself faces issues with "junk information" in its search results, such as thousands of homogenized products and fraudulent reviews appearing when searching for heated ski socks.

Why is Costco Considered an "AI-Proof" Business Model?

Once-rapidly growing Direct-to-Consumer (DTC) brands like Casper (mattresses) and Allbirds (shoes) often failed to build lasting business models. These brands typically did not manufacture their own products, instead relying on OEM manufacturers and then purchasing traffic through Google and Facebook. Once a category proved profitable, commoditization competition quickly led to price wars and margin erosion. Furthermore, the lack of a subscription model and the rapid shifts in fashion trends made it difficult for these brands to sustain growth.

In contrast, Costco is regarded as a unique "AI-proof" business model. Its core strengths lie in:

• Membership Model: Costco's primary profit comes from its annual membership fees, approximately $100, rather than from product sales. This model allows Costco to focus on delivering value to its members.
• Stringent Quality Control and Low Gross Margins: Costco refuses to sell inferior products and maintains exceptionally low gross margins, ensuring product quality and price competitiveness, thereby preserving member value. For example, Costco's Kirkland brand wines, beers, and shirts are renowned, and the company was even sued by Lululemon for producing pants that were superior and cheaper. Its $1.50 hot dog price has remained unchanged for years, and it even built its own chicken farms to control the cost of its rotisserie chickens.
• Consumer Trust: This model has built an extremely high level of consumer trust, with people believing that all products sold by Costco (including its Kirkland brand) are of high quality and reasonably priced.

Alex believes that Costco's business model is difficult to replicate, as it requires decades of trust accumulation, but it possesses strong resilience against AI disruption. It has even been suggested that Costco's board consider entering the financial services sector, offering the lowest-cost loans, which would further enhance member value, with Costco profiting from membership fees.

What New Opportunities and Challenges Will AI Bring to E-commerce?

AI will impact not only consumers but also merchants. Justine notes that merchant websites will need optimization to allow AI Agents to better browse, interact, and make purchasing decisions on behalf of consumers. Simultaneously, new financial infrastructure will be required to support AI Agents in making purchases and using credit cards for consumers. Alex adds that if AI agents control the display layer, Amazon's advertising revenue (derived from ads that direct users away from the Amazon platform) will be threatened, as Amazon would no longer be needed as an intermediary.

Looking ahead to the next decade, the e-commerce sector is expected to see the emergence of new, large, and enduring companies. ChatGPT itself is a new entrant that will play a role in e-commerce. Niche markets focused on those who "value money more than time", such as CamelCamelCamel (Amazon price tracker), Ebates (cashback website), and Quidco (a similar UK cashback company), will be brought mainstream through AI-powered automated price comparison, cashback, and coupon application. AI can make these services incredibly easy to use, thus creating opportunities for startups.

Ultimately, AI companies could emerge as the winners of the "Last Click". Future AI agents may integrate all credit card, cashback, and coupon information, becoming the ultimate executors of purchases. Beyond ChatGPT, Justine believes that specialized AI models, fine-tuned for specific domains (e.g., bicycles), may also emerge, offering a more professional purchasing experience by better understanding user needs.

This AI-driven transformation will compel the e-commerce ecosystem to fundamentally rethink its value propositions, business models, and interactions with consumers.