TABLE OF CONTENTS
- What is a Portfolio Model?
- How is a Portfolio Model Different from Other Models?
- Can you compare portfolio with other models?
- What are the Different Media Effects?
- Explain media effects with an example?
- What is Split View and Combined View?
- What are Aggregated Inputs?
- How many UIDs can I have in a portfolio model?
- How many products can I include in a portfolio model?
- How does complexity scale with the size of the portfolio?
- How does optimization work in portfolio models?
- What is the modeling approach?
- What are the data requirements?
- What is the minimum data requirement?
- What are the Limitations of Portfolio Model?
- How to ask for Help?
What is a Portfolio Model?
A portfolio model is a modeling framework that combines multiple product-level models into a unified structure. It allows for the analysis of media effects across products—capturing not only direct media impact but also cross-product and halo effects. This is particularly useful for organizations with multiple product lines or business verticals that share media investments
How is a Portfolio Model Different from Other Models?
Unlike standalone product models that evaluate media impact in isolation, portfolio models:
- Integrate multiple product models into a single framework.
- Allow for cross-media effects (e.g., ads for Product A influencing Product B).
- Include halo effects from brand-level or non-product-specific media.
- Support cross-pooling of priors across products to improve model stability and interpretability
Can you compare portfolio with other models?
Model | Objective | Scope | Insights Provided |
Direct/Sales Model | Optimize sales and marketing efforts for the product | Understanding factors driving sales for a single product | Detailed insights into product performance |
Churn Model | Reduce churn by identifying factors driving customer attrition | Predicting likelihood of customers leaving | Factors driving churn and strategies for retention |
Brand Model | Optimize brand-building activities | Impact of marketing activities on brand equity and perception | Impact of marketing on brand equity and perception |
Portfolio Model | Understand combined impact of marketing activities | Impact of marketing investments across a portfolio of products | Aggregated and individual product insights |
Long Term Model | Understand short-term and long-term media effects | Sustained impact of media on brand equity and sales | Long-term brand equity and sales impact, media ROI |
Direct-Indirect Model | Understand and optimize both direct and indirect media effects | Measures direct and indirect effects of paid media on sales | Combined impact of direct and indirect media effects |
What are the Different Media Effects?
Portfolio models capture three types of media effects:
- Direct Effects: Media spend directly impacting the KPI of the targeted product.
- Cross Effects: Media for one product influencing KPIs of other products.
- Global Effects: Media not targeted at any specific product but influencing the entire portfolio (e.g., brand campaigns)
- Halo effects: Media of non-modelled product impacting KPI of selected product.
Explain media effects with an example?
Let us consider Apple and its wide range of products.
Advertisements for iPhones not only boost iPhone sales but also positively influence the sales of other products like MacBooks and iPads. This exemplifies cross media effects, where the marketing efforts for one product contribute to increased sales of other products within the portfolio.
Additionally, Apple's brand advertisements, which promote the company's general image, can create halo effects that enhance customer perceptions and drive sales across all its products, from iPhones to Apple Watches to AirPods.
In brief,
- Direct Effects: Impact of marketing activities directly on the product being advertised.
- Cross Effects: Impact of marketing activities on other products within the portfolio.
- Halo Effects: Impact of non-modelled product ads on all products in the portfolio
- Global Effects: Impact of marketing activities of overall brand on the specific product sales.
What is Split View and Combined View?
- Split View: Displays individual product-level results within the portfolio. Useful for understanding how each product performs independently.
- Combined View: Aggregates results across all products to show the overall impact of media on the portfolio. This is helpful for executive summaries and strategic planning
What are Aggregated Inputs?
Aggregated inputs refer to the consolidation of data (e.g., media spend, KPIs) across products or time periods to simplify modeling and improve computational efficiency. This is especially relevant when modeling halo effects or when data granularity is limited
How many UIDs can I have in a portfolio model?
The same limitations apply as for a standard sales model. There is no additional UID constraint specific to portfolio models
How many products can I include in a portfolio model?
Currently, the system supports up to six products within a single portfolio model
How does complexity scale with the size of the portfolio?
Complexity increases rapidly with the number of products due to the exponential growth in cross-media interactions. Each media effect type (direct, cross, halo) is modeled separately to manage this complexity
How does optimization work in portfolio models?
Optimization is performed at the portfolio level, considering the combined impact of media across all included products. The system supports both pooled and unpooled configurations for cross-media effects
What is the modeling approach?
Each product is first modeled individually. These models are then combined into a portfolio model. The modeling platform allows for configuration of priors and pooling strategies across products
What are the data requirements?
Data must be available at the product level for all included SKUs, including media spend, KPIs, and metadata for modeling cross and halo effects
What is the minimum data requirement?
The Portfolio Model requires data at a weekly granularity for a minimum of 3 years, plus an additional 4 weeks for holdout testing.
What are the Limitations of Portfolio Model?
- Product Limit: Typically supports up to six products per portfolio.
- Complexity: Model complexity increases exponentially with more products due to cross-effects.
- Data Requirements: Requires consistent and granular data across all products.
- Interpretability: Cross and halo effects can make attribution more complex.
- Computational Load: Training and validation can be resource-intensive.
How to ask for Help?
- Above the Feature Panel, you'll see a Need Help section.
- You can:
- Write to the support team if you're stuck.
- Browse Search Articles for self-service support and guides.
Note:
- The Model operates exclusively on a weekly basis due to the reliance on brand data, which is only available at a weekly granularity.
- The Media Insights and Business Insights page are dynamic, a change in one section will be reflected in other two sections instantly.