Team Cognition Under Algorithmic Management: Evidence from Distributed Software Development Teams

Algorithmic systems increasingly shape how distributed software teams monitor work and evaluate performance, but their relationship with shared team understanding remains unclear. This study examined perceived algorithmic monitoring and evaluation (AME), supportive leadership, team mental models, and technical task-output performance in 72 distributed software development teams from 32 firms participating in Horizon Europe projects. Survey responses from 317 team members were averaged to team scores and matched with five GTmetrix indicators of each team’s delivered output collected six months later. Averaged perceptions of AME and of supportive leadership were each positively associated with averaged perceived shared understanding. These associations held when the nesting of teams within firms was modeled, but between-team reliability of the aggregated measures was low (ICC(2) ≤ 0.16), so they are associations among team-averaged perceptions rather than evidence of reliable differences in shared team cognition. AME showed nominal associations with two highly correlated technical indicators that did not survive adjustment for multiple testing and were weaker in a web-only subsample of 59 teams that excluded mobile applications measured through store listing pages; the indicators describe page performance under observed deployment conditions. No statistically detectable moderation by supportive leadership was found, and the exploratory, firm-unadjusted bootstrap gave no evidence of a conditional indirect association through team mental models. The study locates the empirical boundaries of a team-cognition account of algorithmic management: perceptual associations are present, but the proposed path to technical performance is not supported in these data.