Ensuring we measure all the nuances of media and non-media factors correctly

We have vast experience in the numerous modelling techniques available in the field of  measurement. We have consolidated this experience and best practices discovered while developing our own proprietary modelling software. This software is designed to support every aspect of our client’s measurements needs and unlock insights and valued recommendations, in an accurate and transparent manor.

Within each model we build, we have two important modelling approach differences that ensures the best in-class media measurement possible for all our clients:

• #1 - We use a multiplicative modelling approach that allows for interactions or synergies between different explanatory variables in the model. For example, an ice cream brand would expect their advertising to generate more sales during hotter holiday periods than colder non-holiday periods. A multiplicative model accommodates this whereas a more basic additive model would not.
• #2 - Given the importance of longer-term advertising effects, we have a proven method for picking these up in our models. We do this by using what we call a dual adstock approach. The model will have media variables for different channels and campaigns, and these will have short-term adstocks with effects that typically last up until 3 months, although sometimes a little longer. In addition, we also use a longer-term adstock for either total spend, or total brand focused spend - this effect would last for usually 3 months to 2/3 years depending on the length of the model.