We describe a dynamic Bayesian model for clickthrough and conversion rates of paid search advertisements in Google for a Dutch online retailer. Understanding the dynamics of ad performance is important for advertisers to make bidding decisions that increase profits. Our contribution to the literature is providing insight into these dynamics. We specify a simultaneous equations model for clickthrough rates, conversion rates, ad position, and the firm’s bid. To study the dynamics, we allow the parameters to be time-varying and control for seasonality. Bayesian results are obtained using a Markov Chain Monte Carlo sampler. Results show that there is indeed substantial time variation in the ad performance. The retailer can use this knowledge to adjust its bidding strategy.