Abstract: Tools and techniques are provided to help estimate the number of unique subscribers to an RSS, Atom, or other online syndicated content feed. Methods, systems, and other embodiments separate feed polling events into groups based at least partially on regularity in their times of occurrence. Grouping of polling events may also be based on client values, such as a client's IP address or user agent identification. Each located group corresponds to one likely subscriber in an estimation of the total number of subscribers to the feed. The use of occurrence times and client header values may be combined with unique URLs, cookies, and other tools to further refine readership estimates.