Answers are grounded in this paper and cite their sources.
We study the problem of dynamically timing shift notifications in on-demand staffing platforms, where experienced casual workers are prioritized through a priority ranking system. Workers are notified of available shifts in priority order, but delayed responses allow lower-priority workers to claim more desirable shifts, causing preference losses for higher-priority workers and undermining managerial efforts to retain a reliable workforce. Managers must also ensure that all available shifts are assigned within a limited planning horizon, despite uncertainty in worker response times. We formalize this trade-off as the Notification Timing Problem (NTP) and establish that even its simplified offline version is NP -complete. To address the dynamic stochastic problem, we propose two threshold-based heuristic policies—the Offline Notification Policy (ONP) and the Expected Buffer Policy (EBP)—derived from offline single-scenario solutions. Using real-world data from an industry partner managing over 15,000 workers across 1.4 million shifts annually, we demonstrate that the proposed policies significantly reduce preference losses and shift vacancies compared with the current notification strategy; in particular, ONP achieves up to a 10% improvement over the policy derived from the stochastic optimization model while reducing computation time from days to minutes.
No takes yet. Share an insight, caveat, or question.
Gawas et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: