Cognitive bias
In agile product and team management, cognitive biases are especially relevant because they affect the quality of estimates, the interpretation of user data, prioritization decisions, and the evaluation of outcomes.
Relevant Biases in Agile Teams
Confirmation bias
The tendency to seek out, interpret, and remember information that confirms prior beliefs. In product discovery: the team conducts interviews looking for confirmation of the solution they already had in mind, instead of looking for evidence that would invalidate it.
Anchoring bias
Giving too much weight to the first piece of data or reference received. In estimation: the first figure someone mentions in a planning poker session anchors the estimates of the rest of the team, which is exactly what simultaneous reveal is meant to prevent.
Availability bias
Overvaluing information that is easy to recall or fresh in memory. In retrospectives: events from the last day of the sprint are remembered better than those from the first week.
Authority bias
The tendency to place excessive trust in whoever appears to be a source of authority. With generative AI: text generated fluently and confidently seems authoritative even when it is incorrect (AI hallucination). In teams: proposals from the most senior person are accepted without question.
Automation bias
Placing excessive trust in automated systems, assuming they are more accurate than human judgment. With AI: accepting generated code without review because "the AI did it."
Survivorship bias
Focusing the analysis on successful cases while ignoring those that failed. In retrospectives: analyzing only what went well without examining the stories that were not completed or that caused incidents.
IKEA effect
Placing greater value on what one has created oneself. In technical teams: resistance to refactoring or discarding one's own code even when the evidence indicates that it is not the best solution.
Biases and generative AI
The widespread adoption of generative AI tools has amplified some existing biases and created new vulnerabilities:
- Automation bias is especially dangerous with AI agents: developers tend to review AI-generated code less carefully than code a human colleague would have written.
- AI hallucinations benefit from authority bias: fluent, confident text triggers the same cognitive response as text from a recognized expert.
Common mistake
Believing that knowing about biases eliminates them. Cognitive biases are largely automatic and unconscious. Knowing that confirmation bias exists does not stop you from falling prey to it in your next user interview. What reduces their impact are structural practices: simultaneous voting techniques to avoid anchoring, interviews with neutral questions to avoid confirmation bias, and cross-reviews to avoid automation bias.
See also
Want to advance in agility? You can search for course and exam calls or go at your own pace by becoming a member of the Agile Club. This membership includes exclusive resources, e-learning classrooms, and access to Skill Arena: a space to practice and measure your agile skills at your own pace.