What is Media Mix Modelling?
Media mix modeling is a statistical analysis technique that processes sales and marketing data to drive better results. It is also known as marketing mix modeling.
The purpose of using media mix modeling is simple - every marketing team should be able to measure the impact of their campaigns. Media mix modeling helps you with just that. Here, the technique uses multivariate regression analysis to compute sales and marketing time series to estimate the trends and impact of the marketing tactics. This in turn helps teams to forecast impacts and set future activities accordingly.
The ultimate goal of media mix marketing is to drive conversion. So, marketing teams optimize their advertising spending and targeting tactics based on the analysis. The can also factor in any external influences to the data like seasonality or promotions to get as accurate results as possible.
How does it work?
Media mix modeling can use both linear and non-linear regression models to determine the impact of ad spending on overall sales. It is essentially a data-driven technique that uses all relevant historic data like - point-of-sales data and other internal data to quantify sales impact.
So, the team has to first establish a relationship between the marketing activities variables and sales variables in quantitative linear or non-linear equations. This is done by portraying the impact of marketing activities on sales volume, the volume generated for each unit of effort (effectiveness), sales generated divided by cost (efficiency), and return on investment.
By creating empirical relationships, the actual time series data can then be analyzed. The ultimate goal is to optimize the variable to maximize conversions. So, an appropriate model is set up where the sales volume/value is the dependent variable and the marketing efforts are considered as independent variables.
This process is of course complicated. Deciphering the marketing variables plus studying which model produces the best output with the least error. Like for any data analysis, the quality of data is also very important. So teams will have to undertake the task of cleaning all the data.
Mostly, a time series of 2-3 years is analyzed at once to factor in the influence of seasonality.
Elements Measured in Media Mix Modeling
This is the list of elements that are all measured when using the media mix modeling technique. Let’s understand more about them.
1. Sales Volume
There are two types of sales volume in consideration here. MMM decomposes total sales into two components, namely - base sales and incremental sales.
Base Sales- This component of sales is driven by base parameters like pricing, long-term trends, seasonality, brand awareness, brand loyalty, among others. These are essentially economic factors that are set and change seasonally.
Incremental Sales- Incremental sales are influenced by and are driven through marketing and sales activities. The total incremental sales are then decomposed into sectors influenced by each marketing activity.
Once we obtain the share of incremental sales in respect to total sales, we understand what part of sales is driven by marketing efforts. That produces deep insight into the effectiveness of the activities being implemented.
Further division gives a more microscopic understanding of the impact of each activity.
2. Media and Advertising
In particular, media mix modeling analyses the sales impact of media and advertising on mediums like television, magazines, and online ads. The MMM results might not produce very definite answers, they can still give valuable insights into how a change in advertising strategies can influence sales.
- We can compare the effectiveness of 15-second vs 30-second execution
- Compare the outcome of running ads on different platforms
- Compare the effect of run times that are during the prime-time vs non-prime time
These insights can heavily affect the ad spending decisions an organization makes, in most cases eventually helping to cut down costs.
3. Pricing
Any change in the pricing structure can directly influence the sales volume. The effect of this change can be seen with the help of MMM. Knowing the percentage of change in sales with respect to the percentage of change in price is essential. Now teams will know the direct impact of price change decisions so they can optimize them better.
4. Distribution
Distribution efforts are a key to successful returns. A well-oiled distribution system can drive growth like nothing else. So MMM, essentially helps teams understand how changing distribution efforts can impact the sales volume/value dependent variable.
The overall holistic understanding of the impact of all distribution channels and their relative costs can help marketers assess the right channels to invest in.
5. Launches
The launch of a new product is meant to drive growth and should in turn have an effect on sales. A successful launch can result in peaked outcomes - which can’t be attributed to the existing variables of the model. For this special variables are needed to capture the incremental effect.
Media Mix Modeling vs Data-Driven Attribution Modeling
Let’s now understand how media mix modeling is different from the data-driven attribution modeling approach.
The data-driven attribution approach is an attribution modeling approach that tracks the engagement of customers by microscopically tracking their customer journeys. This tracking ensures the marketing team spots all the pain points and performs attribution to touchpoints and channels that contribute to a conversion. The evaluation of performance takes place after a few months at the conclusion of the campaign. This follows the process of experience analytics where the customer journey as a whole is at the center of analysis.
The marketing mix model directly analyses marketing tactics over a period of time. The impact of particular campaigns is analyzed. But it further extends itself to understand the influence of seasonality, brand equity, awareness, etc.
The attribution model has a more granular move but can provide results that don’t attribute for offline conversions. On the other hand, MMM is not as granular. It gives a holistic view but fails to provide insights into specific consumer desires.
Startup
