What is Media Mix Modeling?
Media mix modeling (MMM) is an analysis technique that allows marketers to measure the impact of their marketing and advertising campaigns to determine how various elements contribute their goal, often conversion. The insights derived from media mix modeling allow marketers to hone their campaigns based on a variety of factors, ranging consumer trends to external influencers, to ultimately create an ideal campaign that will drive engagements and sales.
MMM uses aggregate data. As such, it is able to evaluate a wider range of channels, both traditional and digital. Additionally, MMM allows marketers to factor in external influencers such as seasonality, promotions, etc.
Techniques Used in Media Mix Modeling
The statistical analysis performed by media mix modeling uses multi-linear regression to determine the relationship between the dependent variable, such as sales or engagements, and the independent variables, such as ad spend across channels. For example, how did increased spend on magazine ads affect overall sales. MMM can use both linear and non-linear regression methods. To get the most robust and accurate visibility into marketing impact, several models should be evaluated.
It is important for organizations leveraging MMM to be discerning when selecting which data they would like to measure and what they are able to measure. Organizations will need to spend time aggregating and cleansing data from internal databases, third-party sources, or both. Media mix models often use two to three years’ worth of data that allow it to factor in items such as seasonality.
The outcome allows marketers to assign numerical value to the impact of campaigns across channels toward achieving their ultimate goal – engagement, conversion, etc. The collection of these insights allows marketers to determine the ROI of their efforts, allocate future spend, and create sales forecasts.
What Does Media Mix Modeling Measure
A good way to understand what media mix modeling measures is to understand why it was created.
MMM came into popular use in the 1960-70s when the marketing landscape was more simplified than it is today.
Kraft was an early user of this type of analysis. As they launched Jell-O, they were able to choose between three or four television networks and magazine advertising to promote the new product.
The approach of traditional MMM allowed them to see if they advertised at different levels - in different parts of the country, at different times of the year - how did that drive sales in those regions. For example, they could advertise Jell-O in ten cities over ten weeks to see if sales increased.
This is MMM in its simplest form, allowing marketers to get high-level insights into campaign effectiveness.
Today as fragmentation has exploded in all of the ways we consume media, MMM data is more often compared to insights from more flexible, granular models.
Media Mix Modeling vs Data Driven Attribution
Both media mix modeling and data-driven attribution models, such as multi-touch attribution, are used to determine the impact of marketing tactics on a business objective. However, it is important to remember that MMM does not examine user-level engagements, such as impressions, clicks, etc. This need for person-level data is why data-driven attribution has become pervasive in marketing. Media mix modeling exclusively measures the impact marketing efforts have on meeting objective, without factoring in the consumer journey.
Data-driven attribution refers to various attribution models, such as multi-touch attribution, that track engagements throughout the consumer journey. These insights allow marketers to understand which tactics have the greatest impact as consumers move down the sales funnel. Attribution models typically evaluate performance after a few months at the conclusion of a campaign.
Media Mix Modeling:
As previously mentioned, MMM provides high-level insights into specific marketing tactics, over a longer period of time. This allows marketers to understand trends such as seasonality, weather, holidays, brand authority, etc. MMM typically analyzes two to three years’ worth of historical data to identify patterns in campaign effectiveness.
Each of these models have uses in modern marketing, but they also both have blind spots. Data-driven attribution models can have limited visibility into offline conversions, and largely focus on digital channels, where MMM can measure both. However, the lack of person-level insight offered by MMM makes it less well suited for customizing campaigns to specific consumer desires.
Why Do Marketers Claim MMM is Broken in Modern Marketing?
As the marketing landscape has become more fragmented with more channels by which to reach consumers, many have claimed media mix modeling is “dead” and does not have a place in modern marketing.
This is because as consumers are exposed to more brand messaging on every channel with which they interact, they have started to tune out messages that are not relevant to their specific needs.
Now, producing ads that do not have an individual in mind can not only reduce marketing ROI, but hurt brand perception in the eyes of the consumer.
As a result, the aggregate insights that MMM provides, which do not delve to the consumer level, do not help marketers to customize messaging to meet consumer demands.
The commonly referred to shortcomings of media mix modeling are:
- Infrequent reports
- Not granular enough data
- Does not look at the relationship between channels
- No insight on brand / messaging
- Does not factor in the consumer experience
The Role of Media Mix Modeling in Modern Marketing
While some believe that Media Mix Modeling is broken, it still has a place in modern marketing, especially when used alongside more consumer-centric models.
MMM is still a simple way to get high-level answers. This analysis can be done infrequently to keep the organizations aware of broad trends and patterns that have occurred over many years.
MMM should not be the primary approach to manage improvements in your marketing strategy, as it is not the best tool to understand how different types of people and messages drive returns. As a result, marketers should not spend a lot on MMM and conduct this analysis once or twice a year. This will provide a historic, high-level diagnostic view on marketing contribution and outside factors interacting with marketing over a long period of time.
Media Mix Modeling Tools
For media mix modeling to be effective today, it must be aggregated with additional marketing measurements, such as multi-touch attribution, to provide a unified measurement. This will give marketers insight into both historical data and person-level engagements with various touchpoints.
This requires a unified marketing measurement platform, that distills big data into actionable insights, for dynamic, in-campaign optimizations.
A unified measurement platform that allows marketers to leverage MMM data alongside analysis from other models should incorporate the following features:
- Provide in-campaign insights
- Offer data integration across all marketing efforts
- Provide granular person-level insights that are informed by historical trends
- Offer analysis on the effectiveness of branding and creative messaging
Additional Resources on Media Mix Modeling
- Changing Your Approach to Marketing Mix Modeling to Achieve Better Results
- Marketing Analytics and the Changing Consumer Landscape
- The Pros and Cons of Marketing Mix Modeling for Enterprises
- Attribution Buyers Guide: How Did We Get Here? A History of Measurement
- Moving From Traditional to Customer-Centric Marketing Mix Modeling