This paper empirically analyzes order flow dynamics in a limit order book (LOB) using high-quality Nasdaq data for five large-tick stocks. The authors study the mean cumulative net order flow at the best quotes before and after market order arrivals. After a market order, they identify four distinct phases: a platform-latency phase with zero net flow, a response-latency phase with slight positive flow, a high-speed reaction phase with strong negative flow at the same-side quote and positive flow at the opposite-side quote, and a lower-speed reaction phase with positive flow at the same-side quote and negative flow at the opposite-side quote. The high-speed reaction phase is attributed to high-frequency traders canceling limit orders, while the subsequent refill phase is consistent with stimulated refill. Before a market order, net flow is negative at the same-side quote, indicating adverse selection avoidance, with a slight reversal just before the order. The findings suggest that liquidity providers balance adverse selection and waiting costs, highlighting the need for multi-mechanism models of strategic liquidity provision.