In the marketing mix, the individual marketing instruments are coordinated and harmonized with each other. But the same question always arises: What is the effect of the individual marketing instruments/activities? Marketing mix modeling can provide this answer.
Marketing mix modeling uses statistical analyses such as multivariate regressions to highlight the influence of the individual factors in the marketing mix. The models show which advertising achieves the best effect on which channels and with which budget.
To measure the impact, the first thing to do is to set a measurable goal or conversion. Ideally, this is done during campaign planning and the campaign is consistently geared towards this. A conversion can be, for example, a sale of a product (sales or number of units), a website visit, a contact, a test drive of a car and/or also a change in the classic advertising KPIs, such as image or advertising recall.
Already during or at the latest after the implementation of the campaign, it is time for the evaluation. First, the relevant data must be collected. This consists of analytics data from the online measures, contact figures from offline media, and the prices of the respective measures. The more detailed the data, the more accurate the evaluation. This data can now be enriched with other influencing factors, such as weather data, competitor advertising data or covid data.
All these data and influencing factors flow into the models. These calculate the influence of the individual campaigns or channels on the defined KPIs, for example on the sales result. For this purpose, a “baseline” is calculated in a first step, which consists of the sales that would have been made even without the marketing activities. The difference then results in the campaign effect.
The exciting question now is which element of the campaign triggered the conversions and how strongly. There are many different methods for this. Signifikants models work with the fluctuations in the media mix over the campaign period on the one hand and the varying campaign effect over time on the other. From this information, the algorithms can determine the effect of the individual channels and advertising media – always taking into account all additional influencing factors outside the campaign, provided that data is available for this.
Intelligent algorithms are a basic requirement for reliable marketing mix modeling. But the whole thing starts much earlier. A data basis that is as clean as possible is the be-all and end-all. This starts with ensuring that all relevant information is included in the models and goes so far as to improve data of insufficient quality by appropriate data cleaning and, where necessary, by enriching it with additional information. In addition, comprehensive plausibility checks are always required at the end. Our experience shows that all these sometimes very time-consuming tasks can be reduced to a very manageable level through automation or clearly defined processes. It also makes sense to verify the results through further analyses of similar campaigns or subsequent flights.
If you take these points into account, you will end up with reliable results and important learnings, thanks to which the efficiency of future campaigns can be significantly increased.