Marketing Mix Modeling (MMM) and Media Mix Modeling (MdMM)

Bidding Optimization

Process of adjusting bids for online advertising in real-time to maximize return on investment (ROI). (Ex: Amazon E-commerce ads)

Key components include:

1. Bid management: Setting and adjusting bids based on performance data.
2. Targeting: Identifying and reaching the most relevant audience.
3. Ad placement: Choosing the right platforms and placements for ads.

The 4 P’s of Marketing

1. Product - What you are selling.
2. Price - How much you are selling it for.
3. Place - Where you are selling it.
4. Promotion - How you are promoting/marketing it.

Marketing Mix Model (MMM)

MMM is a statistical analysis technique used to estimate the impact of various marketing tactics on sales and then forecast the impact of future tactics. It quantifies the ROI (return on investment), and helps allocate advertising budget effectively across channels. (ex: google, facebook, walmart, etc).

1. Estimate the sales function with time-series data (Non-linear function estimation)
   • Sales = $\sum(Budget, clicks, impressions)$ (for each channel)
2. Optimal budget allocation
   • $\max \text{Sales}$

Components of Marketing Mix Modeling

• Dependent Variable: Sales, profit, or market share.
• Independent Variables: Product features, pricing, promotions, place (distribution).
• External Factors: Competitor actions, economic trends, seasonality.

Applications of MMM

• Budget allocation across the 4Ps.
• Price optimization and promotion effectiveness.

Case study

Not sure if this was real or not, but an example given in class.

• large retail chain company
• challenge: sales stagnation despite increased marketing spend
• Solution: MMM identified underperforming promotions and redirected budget to successful channels.
• Outcome: 10% increase with smaller budget

MdMM (Media Mix Modeling)

• A specific subset of MMM that focuses on media channels.
• Evaluates the performance of different media platforms to optimize the media mix.
• Complementary to MMM, often used in conjunction with MMM for a comprehensive marketing strategy.

Components of MdMM

• Dependent Variable: Sales, impressions, brand awareness.
• Independent Variables: Media platforms (TV, online, social, radio).
• External Factors: Audience demographics, time of year, media consumption trends.

Applications of MdMM

• Media budget optimization
• campaign effectiveness by media type

Case study

• Company: Global consumer goods brand.
• Challenge: Underperforming media campaigns.
• Solution: MdMM revealed that TV ads were driving less value than digital ads.
• Outcome: Reallocated 20% of the TV ad spend to digital, resulting in a 15% increase in brand awareness.

Steps to MdMM

Defining the objective and KPIs is important.

1. Define Objectives & KPIs

Gather historical data on sales, marketing activities, and external factors.

• Objective: What is the goal? (e.g., increase sales, boost brand awareness)
• KPIs: Metrics to measure success (e.g., sales lift, ROI, brand engagement)
• Example: A company’s objective is to boost brand awareness with a new product launch. KPIs include impressions, engagement rate, and ad recall.
• Objective: Maximize sales and acquire new customers.
• KPIs: Set clear metrics like sales growth, customer acquisition rate, or return on ad spend (ROAS).

2. Data Collection

• Use regression analysis to identify relationships between variables.
• Gather historical data on media spend and performance across channels:
  • TV: Spend, impressions, reach.
  • Digital Ads: Clicks, conversions, views.
  • Social Media: Likes, shares, engagement.
  • Print: Circulation, audience size.
• Include external factors like seasonality or competitor actions.
• Data Sources: Gather data from various channels like TV, social media, search ads, and in-store promotions.
• ETL Process: Use tools like SQL to extract, transform, and load the data from company databases.
  • automated these days

3. Build a Regression Model

• Quantifies the ROI and sales lift attributed to each marketing activity.
• Use regression analysis to determine the relationship between media spends and the dependent variable (e.g., sales, brand awareness).
• Control for multicollinearity: Media channels often overlap, e.g., TV and online ads.
• Example equation: $$Sales = B0 + B1f1(TV Spend) + B2f2(Digital Spend) + B3f3(Social Spend)$$
• S-Curve: Model the relationship between spend and performance, identifying diminishing returns.
• Other models:
  • diminishing returns model (quadratic + s-curve)
  • hierarchical bayesian model
  • company customized s-curve
  • etc.

4. Analyze Results

• Predict future sales based on different budget allocations.
• Determine which channels contributed the most to your objectives (e.g., ROI by channel).
• Example: TV contributed to 40% of sales, while digital ads contributed 30%.
• Look for diminishing returns: More spending on a channel doesn’t always lead to more results.

5. Reallocate Budget

• Adjust the media budget to focus on high-performing channels.
• Example: Shift budget from TV (with diminishing returns) to digital ads (with higher ROI).
• Continuously monitor and adjust based on updated data.

Metrics

• Return on Investment (ROI): The incremental gain in sales for each dollar spent.
• Media Contribution: The share of total sales or impressions driven by a specific media channel.
• Elasticity: Sensitivity of sales to changes in marketing spend.
• Incrementality: The additional sales generated by a marketing or media activity.

Advantages

• Data-Driven Decisions: Empirical evidence replaces gut feelings.
• Optimized Budget Allocation: Better distribution of marketing spend for maximum impact.
• Actionable Insights: Provides clear, actionable insights for improving future campaigns.
• Long-Term Planning: Helps in developing long-term marketing strategies.

Limitations

• Data Requirements: Large, clean data sets are required.
• Lag Effects: Some marketing activities take time to show their full impact.
• Multicollinearity: Difficulty in isolating the effects of overlapping marketing channels.
• Attribution Complexity: Accurate attribution in multi-channel campaigns can be challenging.

Future Trends

• AI and Machine Learning: More advanced algorithms for predictive modeling.
• Cross-Channel Attribution: Integrating all marketing channels for a unified view.
• Real-Time Data Integration: The ability to analyze and react to data in real-time.
• Incorporation of Social and Influencer Media: New channels like social media and influencer marketing are being incorporated into models.

Walmart AdMix Modeling (WAMM)

• A specific application of MdMM for Walmart’s advertising strategies.
• Walmart leverages in-store and online customer behavior data to optimize ad spend.
• WAMM is Walmart’s proprietary media mix modeling solution, designed to help brands optimize their media spend on Walmart’s platforms.