When expert estimates vary widely, the most effective technique is to facilitate structured consensus-building using a method like Planning Poker or a Delphi-style approach. Having SMEs reveal estimates simultaneously (A), discuss the assumptions behind the highest and lowest values, and then re-estimate reduces anchoring bias and surfaces hidden constraints or missing scope. This improves estimate quality by aligning understanding of work complexity, risks, and dependencies. Using historical information (C) is helpful, but it may not resolve differences if the current work is novel or if SMEs interpret the task differently; historical data should complement-not replace-assumption alignment. Recommending an iterative approach (B) is too vague and does not directly address divergence. Escalating to the sponsor (D) is inappropriate because estimate differences are a normal planning issue best resolved by facilitation and expert discussion, not escalation. Structured group estimation produces more reliable durations, supports buy-in from those doing the work, and strengthens the schedule baseline by making assumptions explicit and agreed.