The utilization of Planning Poker for estimating user stories in the product backlog of an agile team is a favored approach of mine. This technique involves individual estimators presenting their estimates by displaying cards. In the event of discrepancies among estimators, they engage in discussions to elucidate the reasons behind their estimates, seeking clarification from the product owner, who ideally should be present, and persisting until a consensus is reached.

Frequently, team members inquire about the necessity of consensus and whether it would suffice to merely compute the mean of the individual estimates. The predicament with relying on averaging lies in its simplicity. By opting for this approach, teams inadvertently sidestep the invaluable and spirited discussions that are a fundamental aspect of the Planning Poker process. Instead, they tend to engage in only one or two rounds of estimation, followed by a straightforward calculation of the average.

An apparent dysfunction arises when an estimator selects the card representing 100 not because they genuinely believe the task will consume such an extensive timeframe, but rather because they perceive 20 as the appropriate value, while other estimators are contemplating 8 and 13. Consequently, to address this issue and other related concerns, if a team is genuinely compelled to adopt an averaging approach, it would be more prudent to utilize the median (i.e., the middle value) rather than the mean (i.e., the sum of estimates divided by the number of estimates). Engaging in discussions sheds light on numerous intricate aspects that teams overlook when they resort to averaging.

Hence, while my intention is for teams to reach an agreement, the degree of heartfelt concurrence is not of utmost importance. If we reach a consensus on 13, some of us may genuinely believe that it is the accurate estimation. Others might think 8 is the correct value but perceive 13 as being „sufficiently close.“ Still, others may feel that the item under discussion has been deliberated upon for an excessive duration and, even though it should warrant a value of 20, they compromise and settle for 13 merely to conclude the process. Therefore, instead of resorting to averaging, if the team reaches an impasse, I propose engaging in another round of discussions. If the deadlock persists, someone should suggest a reasonable value and assess whether everyone can „support it“ rather than deeming it the „absolutely perfect number.“