There are many reasons why a customer will choose to purchase after clicking a keyword ad. Understanding these factors to accurately predict the conversion rate is key to optimizing an organization's ad buying strategy. However, multiple sources of data are needed across multiple granularities and it is a challenge to aggregate all of that data at the right level to allow for the conversion rate forecast. Marketing often explores performance at a DMA level and it often makes sense to do this modeling at the same level. In this case, we start with historical conversion rates over a period of time (a week) per tem and DMA. This is our target. Next, we want to include information about that term's behavior on the google trend data aggregated to DMA and week. Customer demographics for that DMA can provide valuable insights. Next, external factors that happened leading up to and during that week in the DMA can impact the conversion rate and should be included. These can include historical weather data and information about local events. All this data needs to be cleaned, joined and transformed into valuable ML features before going into model training. This pre-modeling prep process can be frustrating and time consuming. We are here to help.