Determining the right media mix for your company can be complex.
While determining the best media mix for a company was once largely a matter of balancing advertising costs against metrics such as reach and impressions, modern digital marketing makes it much more difficult to do this. In addition to the differences inherent in automated and targeted digital advertising when compared to simply buying an ad slot during a program, there are a number of factors that need to be considered in digital marketing that simply wasn’t an issue with more traditional forms of advertising.
Compounding the issue is the fact that in many cases, you may not have access to all of the data you’d like to have when trying to balance your media mix. This is due to the fact that in many cases the data you’re provided is only an estimate based on average use cases from past advertisers. Fortunately, there is an option available that will provide you with a better picture of your marketing impact so you can more accurately determine your media mix.
Media mix modeling (MMM) can give you the insights you need and cut through the lack of transparency often associated with digital marketing.
What is media mix modeling?
Media mix modeling is a form of statistical analysis that compares factors such as sales or engagements to ad spending across various channels.
This analysis allows companies to see how changes to their ad spend affect overall sales growth or brand strength even when specific metrics for different types of data aren’t available. There is no one specific method used to create these models; a number of different techniques can be used in the analysis, including both linear and non-linear regression techniques.
The strength of MMM lies in the fact that it uses a robust data set and analyzes it to determine trends over time. This allows companies to track multiple trends within the same time period, determining how different factors (such as different types of advertising spending) may have impacted the company’s success within that time. By looking at these different trends, a more accurate media mix can be developed.
How media mix modeling works
Media mix modeling works by taking available data from sources such as internal databases, third-party analytics sources and other sources to develop a large aggregate data source that covers a period of time.
In most cases, at least two or three years’ worth of data is used in developing MMMs to ensure that factors such as seasonal sales variations and short-term economic factors don’t have undue influence over the resultant model. Once the data is collected and cleansed to remove outliers or other potentially disruptive information, statistical analysis techniques are used to chart trends and determine the significance that specific variables (such as digital advertising spending) had on issues such as overall sales, ad engagements or other trends.
In most cases, multiple models will be developed from a single set of data, ensuring that odd results from a single model don’t have undue influence on the eventual media mix for the company. By producing multiple models using different techniques, a clearer view of the actual factors that influence sales and other aspects of the company’s business emerge. This accurate view creates transparency for the company even in situations where direct data isn’t always available.
Applying media mix modeling to your brand
Developing media mix models can take time, but the insight that they provide is certainly worth the effort.
Data must be aggregated from one or more sources, and as mentioned there should be enough data included to reduce or eliminate the influence of short-term factors. Your company also must decide exactly what it is that it’s trying to measure; different variables such as sales, ad reach, click-through rates and the like will need their own calculations, so you need to be as specific as possible.
Once regressions are performed and models are developed, your company’s media mix can be adjusted based on the factors that showed the strongest correlations to sales or other variables that you wish to increase. Keep in mind that additional modeling will be needed in the future, so make sure to reevaluate your needs periodically to ensure that your media mix is still in line with trends in more recent data.