Respondents are randomly selected and monitored to ensure that each sample is statistically representative of the overall adviser population. All respondents are incentivised to respond. Respondents are derived from external sources and are not associated with any publisher or financial services company.

Re-sampling is monitored to ensure that no respondent is duplicated on consecutive surveys. All options for questions are provided in a randomised order to eliminate position skews.

Statistical analysis is conducted on all results to ensure margins of error are kept at a minimum – to do this, rolling averages are used for some data series. The timing and response patterns are carefully monitored to ensure that response fatigue is not a factor in surveys.

The Advertising Effectiveness Reports use split sample analysis to isolate the impact of advertising and other business metrics.

All data is drawn from the monthly tracking surveys. Survey samples are aggregated to ensure a statistically robust sample sizes for each analysis area—No data points are duplicated. The effectiveness report provides insights via correlated (although not necessarily causal) statistical differentials.

No data points are duplicated

All survey data is gathered monthly from practicing financial advisers using an online survey, with an average of 120 for Australia and 230 advisers per month for the UK:

• Publication readership: whether advisers read the publication in the
  previous month.
• Frequency of readership: how many issues in that particular month the
  adviser read.

Six months of data (currently 650 for Australia and 1250 unique advisers for the UK) is accumulated and directly used in the Media Optimiser program after converting it into a binary matrix. When a particular publication combination is selected using the media optimiser program, the full data set is filtered to isolate only those advisers who have read any of those publications.

For non-monthly publications, each adviser’s responses are weighted for inclusion in this refined data set by applying their readership frequency results. For instance any adviser who selects a weekly publication, and who then indicates they read only two of the previous four issues, is weighted as a 50% response media-optimiser

• The weight factors differ for each different publication frequency.
• The factors are derived by dividing the actual readership frequency by the
  maximum readership frequency.

The final weighted data set is then analysed and contrasted with the full data set to derive the reach and frequency statistics. The composition of this final data set is also analysed to determine any demographic deviations from the full sample, which can then be attributed to the selection of the particular publication combination in the filtered data set mentioned above.