#Planning Poker #Estimation #Agile Psychology #Scrum

Why Blind Estimations Work: Eliminating Anchoring Bias in Agile Teams

Discover how hidden voting in Scrum Planning Poker prevents anchoring bias, reduces groupthink, and builds psychological safety in remote engineering teams.

By Giancarlo Girardi • • 4 min read

Why Blind Estimations Work: Eliminating Anchoring Bias in Agile Teams

When agile teams gather to size backlog items in Sprint Planning or Backlog Refinement, a subtle psychological force often distorts the outcome before the real discussion even begins: Anchoring Bias.

In cognitive psychology, pioneer researchers Daniel Kahneman and Amos Tversky demonstrated that when people estimate unknown quantities, their initial guesses are heavily anchored by whatever number is mentioned first. In a traditional meeting without blind estimation mechanisms, a senior architect might casually remark, "This looks like a quick 3." Instantly, the psychological anchor is set. Team members who silently suspected an 8 or 13 will unconsciously adjust their estimate downward to avoid confrontation or appearing slow.

---

The Pitfalls of Open Discussion Before Estimation

Open estimation sessions suffer from three systemic flaws:

1. Seniority & HiPPO Dominance: High-ranking developers or vocal personalities unconsciously set the estimation ceiling and floor, muffling dissenting perspectives from junior or quieter engineers. 2. Bandwagon Effect (Conformity): As consecutive team members declare their estimates out loud, momentum builds toward the emerging majority, hiding critical risks that only one team member spotted. 3. Premature Convergence: Teams rush to agreement on an average number rather than exploring the variance that reveals fundamental architectural ambiguity.

MethodTypical Estimation FlowGroup Dynamic & Outcome
Open Discussion* *(Unprotected)*Senior Dev: *"This looks like an easy 3."* ➔ Dev 2: *"Agreed, 3."* ➔ Dev 3: *"Actually an 8... well, 3 is fine."***Anchor dominates**: Divergent perspectives and critical risks are unconsciously silenced to avoid friction.
**Blind Estimation** *(Teamprove Meet)*Dev 1: [?] (secret) ➔ Dev 2: [?] (secret) ➔ Dev 3: [?] (secret) ➔ **Simultaneous Reveal: [3, 13, 3]***Variance surfaces risk: Hidden architectural assumptions and domain questions immediately come to light.
---

How Blind Estimations in Planning Poker Solve This

The core mechanism of blind estimation—where every team member picks their Fibonacci card privately and reveals them simultaneously—radically transforms the dynamic:

1. Independent Mental Models

Because each participant must submit their vote in private, everyone is forced to run their own internal simulation of the problem, interfaces, and testing overhead. There are no shortcuts through imitation.

2. Variance as a Diagnostic Signal

In Teamprove Meet, the simultaneous reveal is the moment of truth: wide variance is not a failure, but the most valuable diagnostic signal of the refinement. When one developer votes 2 and another votes 13, it highlights that: - The ticket has hidden complexities known only to one person, or - There is a fundamental divergence in understanding the existing codebase or system architecture.

By having the lowest and highest voters explain their reasoning first, the entire team rapidly builds shared context.

3. Fostering True Psychological Safety

Hidden voting democratizes the discussion. A junior engineer's card holds equal visual weight on screen as that of the principal engineer. In remote teams where subtle non-verbal cues are easily lost over video calls, blind voting creates a safe container where honesty is the path of least resistance.

---

3 Practical Tips for Facilitating Blind Estimations

1. Ban Pre-Vote Number Dropping: Clarification questions before the vote should strictly focus on acceptance criteria, never on effort or story points. 2. Celebrate Outliers: Treat wide divergence as a gift of risk prevention before sprint commitment, not as conflict. 3. Use Strict Visual Timers: Keep the cadence brisk. A 45-second timer prevents over-analysis and encourages intuitive sizing based on experience.