The paper applies Kalman Filtering to investigate dynamic relationship between electronic word-ofmouth (e-WOM), online advertising and offline product sales. We explore three dimensions of e-WOM: volume, valence and variation. We construct online advertising in terms of page view of online display advertisement. Using a merged, multi-sourced, time series dataset, we reveal that all aspects of e- WOM are significantly related to sales. In initial stage, e-WOM volume takes the dominant effect in driving the sales, while e-WOM valence becomes more effective later on. And more dispersed opinions lead to higher sales. However, counter-intuitively, online advertising is not significantly related to offline sales. The paper enriches our understanding of the link between online marketing and offline sales.