The need for the evaluation of advertising is nothing new, but most of the processes used currently are typically based on establishing which media to use, not on what to say to whom. There is no reason why modelling cannot robustly work across how, when, who and what to say.

In some cases advertisers will use simple share of voice models that work on the basis of it is not what you say to who and when, but how much money you spend on doing it. Econometric modelling is more demanding, but is a far more robust way of planning advertising expenditure and effort.

Many will be familiar with the power of econometric modelling. True, the use of econometrics has many limitations, being a reflective process predicting future behaviour in a time where factors such as the macro economy and consumer behaviour are subject to sudden and significant change. In the last few weeks we have seen the threat of increased interest rates, increasing VAT, and the threat of another oil crisis like that of 1973, but most models will have intervention mechanics that will let agencies and clients chuck in extraordinary events, typically using the power of their imagination to create the variables.

But a self imposed limitation on the use of modelling is to use its outputs solely for optimising media mix, when the mix of messaging being carried by that media is often where there is the greater imbalance is to be found. It is not about how many TVRs you might be using on television, but how many are used for building the brand and supporting other media rather than how many are used for direct response. Very few models ever examine this.

Download full document here.