Marketing Media Mix Modeling Examples: A Practical Guide for Smarter Budget Decisions
Overview
Rather than spending a ton of money on advertising and praying that one of these will turn out to be a winner; today’s marketeers are able to use a more scientific method which can be compared to cooking. In order to prepare a great dish, the right amount of seasoning must be added, otherwise it could end up too salty or too bland.
Media mix modeling (MMM) allows the marketer to accurately measure the impact of each media type to see how they each fit into the overall results. Marketers are using media mix modeling (MMM) as an easy to understand method. Let’s take a look at some real examples of how different brands are leveraging MMM in order to make more accurate decisions.
What is marketing media mix modeling?
Media mix modeling is a way to analyze how various marketing channels impact business results such as revenue, sales, or leads. Think of it as a fitness tracker for your marketing budget by determining how effective specific marketing channels were based on previous historical data to show success rates.
Some things that are analyzed are:
-The amount of money spent on advertising
-How each marketing channel was performing
-The amount of sales (outcomes/results)
-Any external environmental events that may have affected ads (seasonality, pricing changes in the economy, etc.)
-Promotional and discount programs
After analyzing all of these factors, the program will finally show the firm what exactly creates real change in the company’s bottom line.
Importance of Media Mix Modeling in Today’s Marketing World
The number of ways to market is endless:
Search Marketing
Social Media Marketing
Television Advertising
Influencer Marketing
Email Marketing
Digital Display Advertising
Sponsorships
Podcast Advertising
But how do you know which of these marketing channels actually work to drive revenue?
Media mix modeling uses numbers to answer your question—not subjective opinions.
Media mix modeling will allow you to:
- Eliminate excessive spending on low-performing channels
- Reinforce spending on high-performing channels
- Predict future results
- Confidently develop your budget
The Media Mix Model Functions Step-by-Step
Data Collection Types
In summary, data will come from many different sources:
Advertising Expense Detail
Advertising Campaign Report
CRM Data
Sales Data
Website TraffUnderstanding output
Data represents more than simple charts; they will help you drive decisions.
You can establish:
1) Channel contribution %s
2) ROI by media type
3) Saturation points
4) Optimal spend levels
In a sense, these analyses convert fog into a clear picture.
Benefits of Media Mix Modeling
Why do large brands favor MMM?
Because it provides:
1) Budget clarity
2) Channel ROI insights
3) Ability to forecast
4) Risk reduction
5) Long-term planning capability
MMM is even more valuable when you have limited ability to track at the user level.ic Analysis
Market Trend Analysis
The more comprehensive the data source, the more reliable the model becomes.
Statistical Modeling Process
Statistical regression analysis models how each marketing input relates to business output.
After constructing the model, the result is a visual representation of the relationship between the various spending levels and total results generated over a specific period.
As analysis is done, patterns are discovered; therefore, there may be some surprises also.
Understanding output
Data represents more than simple charts; they will help you drive decisions.
You can establish:
1) Channel contribution %s
2) ROI by media type
3) Saturation points
4) Optimal spend levels
In a sense, these analyses convert fog into a clear picture.
Benefits of Media Mix Modeling
Why do large brands favor MMM?
Because it provides:
1) Budget clarity
2) Channel ROI insights
3) Ability to forecast
4) Risk reduction
5) Long-term planning capability
MMM is even more valuable when you have limited ability to track at the user level.
Types of Media Mix Models
Offline Media Types
Television
Radio Stations
Print Media
Billboards / Outdoor Advertising
Offline Channels are More Important Than People Think:
Digital Media Types
Google Ads
Social Ads
Display Ads
Email Marketing
Affiliate Programs
Digital Provides Quick Feedback but Poor/Noisy Quality Signals.
Channels
Influencer marketing campaigns
Sponsorships
Event marketing
These usually have delayed yet significant effects on your business.
Real-World Marketing Media Mix Modeling Examples
Now let’s put theory into practice.
Example 1: Retail Brand Budget Optimization
One national retail franchise used MMM to study how and where to spend over the two years of
expenditures.
What they thought:
That social media was their primary source.
What the model calculated:
That 42% of all sales were from TV advertising.
What they did:
They reallocated 18% of their budget back toward TV, resulting in an 11%
increase in total sales.
Takeaway: Trust data over intuition.
E-commerce ROAS enhancement example #1: For an online retailer, ads were deployed across search, social and display channels.
Findings from the MMM suggested:
- The fastest conversion rate was achieved through search ads.
- The most significant influence on assisted conversions was derived via social media.
- The use of display media generated the awareness. However, the ROI from display media was low.
As a result, KPIs were adapted depending on the specific role of each channel.
As a result, the returning on advertising spend improved by 27%.
All channels are not necessarily closing sales (Some Open Doors)!
E-commerce ROAS metrics example #2: For a manufacturer of consumer goods, digital was thought to be replacing TV.
Findings from the media mix modelling indicated:
- TV created short peaks of demand.
- Digital captured that demand.
- Together, the combination of both channels achieved more than either channel on its own.
The advertising budget strategy was amended to synchronize campaigns in order to enhance sales.
As a result, the resulting sales lift is 19%.
Advertising channels are not competitive but rather complimentary!
E-commerce ROAS metrics example #3: A technology branded company launched a new gadget with the following channels:
- Influencer TM advertising
- YouTube advertising
- Paid Search
- Work with public relations agencies.
Based on the MMM analysis, it was determined that the majority of consumer purchases occurred through influencer TM advertising in combination with retargeting through YouTube advertising.
Because of this, subsequent launches were designed with an influencer-first sequencing strategy.
As a result, launching products through influencer-first sequencing resulted in dramatic improvements in launch efficiencies.
E-commerce ROAS enhancement example #1: For an online retailer, ads were deployed across search, social and display channels.
Findings from the MMM suggested:
- The fastest conversion rate was achieved through search ads.
- The most significant influence on assisted conversions was derived via social media.
- The use of display media generated the awareness. However, the ROI from display media was low.
As a result, KPIs were adapted depending on the specific role of each channel.
As a result, the returning on advertising spend improved by 27%.
All channels are not necessarily closing sales (Some Open Doors)!
E-commerce ROAS metrics example #2: For a manufacturer of consumer goods, digital was thought to be replacing TV.
Findings from the media mix modelling indicated:
- TV created short peaks of demand.
- Digital captured that demand.
- Together, the combination of both channels achieved more than either channel on its own.
The advertising budget strategy was amended to synchronize campaigns in order to enhance sales.
As a result, the resulting sales lift is 19%.
Advertising channels are not competitive but rather complimentary!
E-commerce ROAS metrics example #3: A technology branded company launched a new gadget with the following channels:
- Influencer TM advertising
- YouTube advertising
- Paid Search
- Work with public relations agencies.
Based on the MMM analysis, it was determined that the majority of consumer purchases occurred through influencer TM advertising in combination with retargeting through YouTube advertising.
Because of this, subsequent launches were designed with an influencer-first sequencing strategy.
As a result, launching products through influencer-first sequencing resulted in dramatic improvements in launch efficiencies.
Illustration 5: Seasonal Sales Forecasting
A beverage company utilized the marketing-mix model (MMM) using seasonality and weather data.
Insight from Their Model
Sales were more reliant on temperature fluctuations versus advertisement expenditure.
Radio advertisements were most successful during hot periods.
Radio advertisements were able to trigger bursts of advertisement activity based on the weather.
Outcome
They had better timing to reduce waste (excessive spending) on advertising.
Media Mix Modeling (MMM) vs Attribution Modeling.
People use these two terms interchangeably too often.
Let’s differentiate these terms.
Differences between Attribution Modeling and MMM
Attribution Modeling
User-specific as it relates to visit history (click-path)
Focused on short term (i.e. day/week) quantitative data
Primarily digital in nature
MMM
Aggregate level data
Long-term impact on sales
Utilizes a combination of offline and online data
Statistical and mathematical models are used
When to Use Attribution versus Media Mix Modeling (MMM)
Use Attribution when:
You have some level of user tracking information
You need to optimize “in-flight” advertising tactics
Use MMM when:
You have a need for strategic budget planning
You cannot track users due to privacy restrictions
You have some form of offline advertising budget to account for
The most effective approach would be to implement both.
Examples of Popular Tools Used for Media Mix Modeling:
Google Lightweight MMM framework
Regression Models in Python/R
Bayesian Modeling Platforms
Specialized Analytics Providers
Custom Models => outperforming pre-made templates.
Difficulties in Media Mix Modelling
MMM is effective – but not magic.
Quality of Data
If data is poor, then the model will yield poor results.
In the case of missing sales data or invoices not being included into a database, accurate models are unable to be developed.
Channel Overlap
Channels may affect other channels.
To separate the effects of different channels is challenging but possible.
Delayed Effect
Certain ads can take longer to create a reaction.
Models will also need to incorporate time periods when calculating the effectiveness of advertisements in order to avoid undervaluation.
Best Practices for Developing Effective Media Mix Models
Utilize Clean Data
Use standard naming, timing and accounting methods.
When utilising a clean data strategy, give preference to data consistency instead of large data sets.
Update Continuously
Once a quarter do a refresh of your model to ensure your model stays current with market trends.
Collaborate Across Teams
To be successful at MMM, your Finance, Marketing and Analytics departments need to work together.
MMM is a team sport.
The Future of Media Mix Models
There are stringent Privacy laws being enforced with the elimination of Third Party Cookies.
MMM will be an alternative to personal or identifiable tracking and is becoming the most popular means of measuring the effectiveness of media.
Expect to See
AI Built Models
Speedy Simulation Scenarios
Real-time Budget Optimisation
Predictive Models vs. Estimate Models.
Summary
Marketing Media Mix Modelling Examples confer that intelligent Brands always measure before moving. You have a true account of what your Marketing efforts have done and can continue to do through proper analysis. By using data to guide their strategies (not chasing “shiny objects”), these Brands utilize MMM to develop marketing strategies that produce calculated ROI–an essential tool to achieving better ROI, generally more accurate predictions, and greater confidence.
Frequently asked questions(FAQs)
1. Which marketing channels in your current strategy do you think are overfunded or underfunded — and why?
2. Have you ever used data modeling or analytics to reallocate your marketing budget? What happened?
3. If you could measure the true ROI of one channel today, which would you choose first?
