10 Scrum Poker Best Practices for Better Estimates
Proven techniques to make your sprint planning sessions more accurate and efficient.
1. Estimate Effort, Not Time
Story points measure relative complexity, not hours. A "5" means "about twice as complex as a 3" — not "5 hours of work." This mindset shift is the single most important thing for accurate estimation.
2. Use a Reference Story
Pick a well-understood, recently completed story as your "reference 3" (or whatever your baseline is). Before each session, remind the team: "Remember, the login page redesign was a 3. Estimate relative to that." This creates consistency across sprints.
3. Discuss Before Voting
The Product Owner should briefly explain each story before voting. A 2-minute overview of acceptance criteria prevents misunderstandings that lead to wildly different estimates.
4. Vote Simultaneously
Never reveal estimates one by one — this creates anchoring bias. Use a tool like Scrum Poker Online where everyone votes privately, then all cards are revealed at once. This is the core principle of Planning Poker.
5. Discuss the Outliers
When estimates diverge significantly (e.g., one person says 2, another says 13), ask the highest and lowest to explain their reasoning. Often, one person knows about a hidden complexity or the other is overlooking a requirement. This discussion is where the real value of planning poker lives.
6. Re-vote After Discussion
After outlier discussion, do a second round of voting. Estimates usually converge. If they still diverge after two rounds, go with the higher estimate — it's safer to overestimate than underestimate.
7. Keep Stories Small
If a story is estimated at 13 or higher, it's probably too big. Break it down into smaller stories. Large estimates are inherently less accurate. Aim for stories that are 1-8 points.
8. Use ? and Coffee Cards
The "?" card means "I don't understand this story enough to estimate." The coffee cup means "I need a break." Both are valid and important signals. Don't pressure people to give a number when they genuinely don't know.
9. Time-Box Each Story
Set a timer for 3-5 minutes per story. If you can't reach consensus in that time, mark it for follow-up and move on. A 2-hour sprint planning session that only estimates 5 stories is a waste of everyone's time.
10. Track and Review Accuracy
After each sprint, compare estimates to actual effort. Over time, patterns emerge — maybe your team consistently underestimates backend tasks or overestimates UI work. Use the dashboard stats in Scrum Poker Online to spot these trends.
Try it free — no sign-up required
🃏 Start Scrum Poker