Methods, systems, and computer program products are provided for measuring effectiveness of content items included in a campaign. One example method includes providing a corpus of campaigns, where a campaign defines parameters for presentation that specify circumstances under which content items are presented to users in response to requests. The corpus is filtered by identifying one or more candidate campaigns, where each of the identified candidate campaigns is associated with a geographic region and includes one or more constraints that permit pausing distribution of content items without compromising campaign goals. For each candidate campaign, one or more experimental geographic locations included in the geographic region are determined that are substantially representative of the region. Presentation of at least one of the content items is paused for a time period, and content item effectiveness is measured by comparing resource conversions for experimental and control geographic locations.