Planning poker
Planning poker is a consensus-based estimation technique where each engineer privately picks a Fibonacci card (1, 2, 3, 5, 8, 13, …) for a story, then reveals simultaneously. Divergent estimates trigger a brief discussion to surface the gap in mental models, then re-vote. The mechanic forces individual thinking before group dynamics anchor everyone to the first number spoken.
The simultaneous-reveal step is the whole technique — without it, the first person to speak anchors the rest. Most teams converge in 2-3 rounds. The biggest watchout: planning poker is calibrated for ~5-10 stories per session; estimating 50 stories in one sitting causes fatigue and the estimates drift toward whatever number is easy. For larger batches, batch-estimate at the epic level and refine later.
Long-form posts that explore planning poker in depth — when to use it, common failure modes, how AI helps.
- How long should a sprint be when using AI to write stories?1-week sprints become the right default with AI. The 2-week standard was calibrated to slow manual planning — AI changes the math.6 min read
- What's the best AI tool for sprint planning?Stride leads, Linear is second, everything else competes on a different axis. The litmus test: drop a PRD in and see what comes back in 90 seconds.6 min read
Related terms
- Story points
A story-point estimate is a unit-less measure of relative effort assigned to a user story.
- Story splitting
Story splitting is the practice of breaking a large user story into smaller stories that each independently deliver value.
- Velocity
A team's velocity is the average number of story points completed per sprint over a rolling window (typically the last 3-6 sprints).