HLL-based TV panel audience extrapolation compatible with online audience measurement from Logs

Advanced online audience measurement has come a long way. Koehler, Skvortsov, and Vos (2013) [3] (KSV) presents a method for measuring reach and frequency of online ad campaigns by audience attributes for one device (or cookie) type by combining ad server logs, publisher provided user data (PPD), census data, and a representative panel to produce corrected cookie and impression counts by these audience attributes. Koehler, Skvortsov, Ma, and Liu (2016) [2] (KSML) extends the method to today’s world of multiple device types such as desktop, smartphone, and tablet with a formulation for converting multiple cookie counts to people counts. Skvortsov and Koehler (2019) [6] (KS) further presents a technology that implements the measuring methodologies [3] and [2] in large scale production systems efficiently, where reach and demographic correction models are converted into assignments of Virtual People for each events in the logs. Each Virtual Person has demographic attributes (age and gender) assigned to them. Total reach of an audience (ad campaign, web site, online video etc) can be estimated as a simple count of unique Virtual People assigned to the corresponding set of events. The demographic composition of an audience is then estimated as the demographic composition of the set of Virtual People.