The Roy Morgan pixel captures a census of digital events for any website or ad campaign where the pixel has been placed.  Based on the information collected from the event, the unique audience model will treat the event in one of two ways.

For events generated by the 19+ million device panel, an algorithm is applied to convert the number of devices to the number people.  This algorithm takes into account the number of impressions generated from each household in the panel and the average number people per household in the panel visiting any given website, as well as demographic information such as Helix Personas.

For the events collected off panel, a different model is applied.  This model leverages the information already gathered about known panelists, as well as information from Roy Morgan Single Source (e.g. website data, device usage, HH composition, matching factor) to impute the known information about the events, to calculate the unique audience from the impressions generated outside the panel.

The results from both models is aggregated together to produce unique audience figures, as well as impressions, for the campaign and websites being measured.The proportion of events generated from known and unknown users will vary for each website or campaign. However, on average at least 40% of events are from known devices.