The Website as a "Stage Set" - Major Transformation of E-commerce and Marketing in the Era of Generative AI

1. Introduction: Declaration of the Web's Main Role Change

"Websites are nothing more than stage props." This statement by iBusiness journalist Dominik Grollmann sent ripples through the European tech industry. The day after its release, marketers and developers took to LinkedIn to express their opinions in long posts, and on X (formerly Twitter), the hashtag <#WebsiteAlsKulisse> briefly trended.

"The era of showing all UI is over. Conversations suffice." — @tech_oliver
"If we only follow AI-recommended 'best products,' price competition will become even more intense." — @ecom_julia

The article suggested a near future where generative AI takes on the role of 'salesperson,' and HTML's main stage becomes the backstage.ibusiness.des4-experts.com

2. The Rise of AI Gatekeepers

Even just looking at the first half of 2025, conversational AIs like Amazon's Rufus, Shopify's Sidekick, and Google's AI Overviews have been redefining the entry points for search and e-commerce. Users can express their desire to "buy a 27-inch monitor that supports 4K at 60Hz for under 50,000 yen right now" in natural language, and the AI will cross-reference inventory, prices, reviews, and delivery dates to provide a single "optimal solution"—not dozens of links.

Traditional sites are relegated to roles like "detailed display of purchase options," "alleviating customer concerns," and "staging the brand's world," and do not stand at the entrance. Grollmann's term "Kulisse = backdrop" suggests this structure.

3. From SEO to AIO: New Optimization Metrics

AIO (AI Optimization) is an initiative to optimize how AI evaluates product information and incorporates it into responses. Instead of chasing search rankings,

1. Structured Data: Explicitly state attributes using Open Product Data and GS1 standards.

2. Vocabulary Design: Publish brand tone as a style guide for LLMs.

3. Dynamic Feeds: Provide inventory, pricing, and local information via real-time APIs.

4. Context Score: A metric to measure trustworthiness when AI cites information (also known in the industry as the "Trust Layer").

The European E-commerce Association (Ecommerce Europe) has announced that it will officially recommend "AIO-compliant metadata" in its 2026 guidelines.

4. Case Studies

4-1. German Furniture Retailer OTTO

Converted their headless CMS into a GraphQL API and published their catalog via a ChatGPT plugin. After transitioning to AIO, the sales ratio via AI increased from 14% to 37%.

4-2. UNIQLO Japan

The "UNIQLO Coordination AI" integrated into their app interacts with customers' closet photos to make suggestions. While the average time spent was halved, the number of purchases increased by 1.3 times.

4-3. Small D2C Brand "Organic Loop"

Heavily reliant on Shopify Sidekick, their acquisition costs plummeted. However, a change in AI's recommendation logic led to a 40% drop in sales for a week, highlighting the "algorithm risk."

5. Temperature Differences on Social Media

While the "no need for official websites" theory is gaining momentum on X, there are also concerns on Threads about the dangers of entrusting brand experiences to AI.

• Proponents (about 62%):

  • "Like catalog shopping in the 80s, AI will just become a 'paper substitute.'"

  • "With no need for UI, accessibility will improve dramatically."

• Skeptics (about 38%):

  • "AI recommendations impose the 'greatest common divisor of the majority.'"

  • "The brand's voice becomes diluted, leading to mere comparisons of price and inventory."

On TikTok, a project video called "#ChatPurchaseChallenge" is spreading. It showcases users completely relying on AI to make purchases, with cumulative views reaching 420 million.

6. Legal and Ethical Issues

1. Transparency of Display Order: The EU DSA amendment (scheduled for implementation in 2026) is being discussed to mandate the disclosure of recommendation logic for conversational AI.

2. Data Monopoly: The purchasing intent data acquired by platformers may fall under "specific category information" under GDPR.

3. Creator Compensation: Rights management for reviews and blogs cited by AI remains undeveloped.

7. Developer Perspective: Toward Headless + LLM "Quasi-Native" Design

In the developer community, a design guideline called "LLM-First Architecture" is beginning to be shared. Its features include:

• Making all UI elements generatable from APIs, allowing direct calls by AI

• Adding semantic information with JSON-LD/Schema.org extended metadata

• Incorporating Prompt Engineering into the CI/CD pipeline, managed via GitOps

On GitHub, PRs including "prompt-tests.yml" have increased 4.8 times year-on-year. This statistic symbolizes an era where AI inference becomes part of the UI.

8. Three Actions Brands Should Take

1. AI Persona Strategy: Design the "first-person" narrative when AI speaks for the brand (e.g., sincerity/playfulness/expertise).

2. Acquisition of Zero-Party Data: Collect unbiased attribute data through direct dialogue with customers before passing it to AI.

3. Multimodal Experience: Expand sites that have become "backstage" into experiential channels through voice, AR, and IoT integration.

9. Conclusion: Competition Begins "Outside the Screen"

The future Grollmann envisions is a paradigm shift where the center of the web moves from "pages on the screen" to "AI's conversational space." However, the backstage is not becoming unnecessary. Rather, it is becoming an era where the brand's philosophy and data quality are questioned in unseen places.

"The backstage is the true protagonist." Only companies that reinterpret this will be able to shine in the spotlight on the new "stage."