---
id: "claim-instant-checkout-failure"
type: "claim"
source_timestamps: ["§ The Habit Playbook"]
tags: ["case-study", "product-failure", "ecommerce"]
related: ["concept-re-completion-rate", "entity-openai", "action-optimize-second-transaction"]
confidence: "high"
testable: true
speakers: ["Yuanyuan Gina Cui", "Patrick van Esch", "Jan Kietzmann"]
enrichment_status: "unverified — no external documentation of the specific product/timeline"
sources: ["attention"]
sourceVaultSlug: "hbr-seg-attention"
originDay: 4
articleStem: "hbr-tier2-07-chinese-ai-firms-habits"
sourceUrl: "https://hbr.org/2026/06/lessons-from-chinese-ai-firms-on-owning-customers-habits"
sourceTitle: "Lessons from Chinese AI Firms on Owning Customers’ Habits"
---
# OpenAI's Instant Checkout failed by optimizing for the first transaction

## Claim: Instant Checkout failed by winning the first transaction, not the second

**Confidence: high (per authors) · Testable: yes**

[[entity-openai-d7]] launched **"Instant Checkout" in September 2025**, letting users buy products from **Shopify and Walmart** directly within ChatGPT, **subsidized with free premium access**.

The authors claim it **failed** — abandoned in **March 2026 with only ~30 Shopify merchants integrated** — because OpenAI optimized for the **first transaction (discovery)** rather than the **second transaction (habit)**. Users discovered products in ChatGPT but **reverted to existing habits** (Amazon or Walmart's own apps) to actually complete and *repeat* purchases.

This is the canonical failure illustration for the [[concept-re-completion-rate]] and the rationale for [[action-optimize-second-transaction]]. It also demonstrates why subsidizing *access* rather than *behavior* (contrast [[action-subsidize-behavior]]) does not build a [[concept-habit-moat]].

**Enrichment / external validation:** ⚠️ **Unverified.** There is **no external documentation** of an OpenAI product named "Instant Checkout" with this Shopify/Walmart scope, launch (Sept 2025), or abandonment (Mar 2026) and merchant count. It appears to be either an internal/experimental program not widely covered, or an **illustrative case constructed by the authors**. The underlying *strategic diagnosis* — many AI-commerce experiments over-optimize discovery/novelty instead of habitual repeat purchasing — is well aligned with known e-commerce and behavioral-design patterns.