Distribution, Defaults, and Switching Research Note

Distribution, Defaults, and Switching Research Note

Distribution should be analyzed as a chain: reach, permission, activation, successful task, repeated use, workflow integration, renewal, and expansion. A large installed base establishes only the first part unless the company discloses or independent evidence supports later stages.

Default and status-quo research shows that existing or preselected options can affect choice. That evidence does not prove that every software default wins. A large capability, price, policy, reliability, or trust difference can overcome inertia.

Switching cost includes discovery, evaluation, procurement, security review, contract work, data migration, integration changes, prompt and workflow redesign, retraining, lost history, monitoring changes, and operational risk.

The European Commission and UK Competition and Markets Authority have documented cloud interoperability, portability, egress, licensing, and multi-cloud barriers. These are useful concrete examples of platform switching issues without assuming that every AI application has the same lock-in.

Google's 2025 Gemini 3 enterprise launch and 2026 Agent Platform launch show distribution through Gemini Enterprise, Vertex AI, developer tools, the Google Cloud console, and an expanded model garden. Anthropic reports broad access across AWS, Google Cloud, and Microsoft Azure. OpenAI describes consumer reach as a channel into enterprise.

These are first-party channel descriptions. They do not independently establish activation, retained use, paid conversion, or exclusive adoption.