Dynamic group coordination facilitates adaptive division of labor in response to group-level changes. Yet, little is known about how it can be operationalized in online collaborations among individuals with limited information about each other. We hypothesized that simple social information about the task distribution of others can elicit emergent task allocation. We conducted an online experiment where participants analyze images of a polluted canal by freely switching between two tasks: creating keyword-based tags for images and categorizing existing tags. During the task execution, we presented experimentally manipulated information about the contrasting group-level task distributions. Participants did not change the effort allocation between the tasks when they were notified that the group deficits workers in the task they intrinsically prefer. By contrast, they allocated more effort to the less preferred task than they would intrinsically do when their intrinsic effort allocation counterbalances the current distribution of workers in the group. Such behavioral changes were observed more strongly among those with higher skills in the less preferred task. Our results demonstrate the possibility of optimizing group coordination through design interventions at the individual level that lead to spontaneous adaption of division of labor at the group level. When participants were provided information about the group-level task distribution, they tend to allocate more effort to the task against their intrinsic preference.