The Roadmap Is a Lie

Petros Bountis, a director of product engineering, proposes five written questions that every roadmap item must answer before engineers are asked to estimate it.

Petros Bountis is a director of product engineering who writes The Socratic Leadership, a newsletter on engineering leadership. He argues that roadmaps present untested ideas as firm promises, and that AI makes this worse because a rough idea quickly becomes a polished prototype and plan. His example is invented: an invoicing feature that took two teams most of a year and was used by four accounts.

• The questions cover the customer behaviour that started the idea, the change it should make and the single belief it rests on. They also ask what cheap evidence would prove it wrong and what result would stop the work.
• The roadmap shows obligations, such as contracts and regulatory deadlines, with dates, and shows untested bets in a different style, with a funding window and a condition for stopping.
• Before the review, the product manager answers the questions in writing. The engineering manager adds a cost range, the effort to verify the work and how easily it could be undone.
• Each bet leaves the review marked to invest, harvest, watch or exit, with a decision owner, a budget and a review date.

A product and engineering team can add the five questions to its existing roadmap review for the largest items waiting for capacity. The author describes the method without results from a team that has used it.

Derived by Working Surface from the article; more in the Playbook. Source line: Coherence has become cheap. Evidence has not.

18 October 2026: as AI agents make code and prototypes cheap, teams are writing down which work needs a human check and which ideas have earned more effort.