A proof of concept (POC) tests whether a solution’s core assumption holds — technically, operationally, or commercially — in a bounded experiment before major spend.

How it works

Parties agree upfront on scope, timeline, success metrics, and who pays. A POC might run four to twelve weeks: integrate with one API, process a sample batch, or hit a latency target on representative data. It is narrower than a paid pilot or production rollout. Enterprise sales often require a POC before procurement approves a multi-year contract.

For investors, POC traction with a named customer reduces technical risk in seed and Series A memos. The distinction matters: a demo in a sandbox is marketing; a POC uses the buyer’s environment and passes agreed gates.

Why it matters

• Founders: Price POCs thoughtfully — free POCs train buyers to never pay; overly heavy POCs burn engineering without a path to contract.
• Investors: POC completion rates and conversion to paid pilots signal sales and product maturity.
• Operators: Write exit criteria so a failed POC ends cleanly instead of dragging into endless custom work.

Common mistake

Calling a sales demo a POC. Without the customer’s systems, data, and signed success criteria, you have a presentation, not proof.

Related ideas

Pilot programs, /glossary/technical-diligence, MVP, and design partners.