Skip to content

Latest commit

 

History

History
37 lines (25 loc) · 3.69 KB

README.md

File metadata and controls

37 lines (25 loc) · 3.69 KB
Raw

Marketing Strategy Optimization

This project was a part of MIT's course 15.093 Optimization Methods.

Purpose and Overview

The project aims to design a framework for optimizing a marketing strategy given a fixed budget, multiple marketing options, varying constraints, and optimization goals. Four marketing options are included here: Targeted Online Advertising, E-mail, Print Media, and Influencer Marketing. However, the frameworks and formulations we provide can be easily extended to fit any combination of marketing options.

Data

The data for this project is synthetically generated. We have four different marketing options for a product: Targeted Online Advertising (max investment value: $990), E-mail (max investment value: $500), Print Media (max investment value: $700), and Influencer Marketing (Social Media) (max investment value: $600). For each of these options, we have the following variables in the data:

  • Investment Amount: Each row defines an investment amount that generates an expected number of views and purchases made.
  • Number of People Reached: Number of people who view the marketed advertisement
  • Sales Generated: Number of people who purchase the product after getting influenced by the marketing through the given platform.

For Data Pre-Processing: We interpolated the data to create a continuous function from our discrete data points, allowing the optimization model to work effectively with any investment amount within the provided ranges.

Methodology

  • Two Baseline Models with only budgetary constraints with the following objectives: (a) Maximizing Views (b) Maximizing Sales

  • Mixed Integer Optimization

  • Additional Constraints: fixed budgetary allocation to specific marketing options, if-then constraint, balanced portfolio constraint

  • Dual: The dual interpretation of our project is also a useful framework to consider as we begin adding constraints to our problem. Below we take the fixed allocation constraint example and extract the dual values from our model.

Graphic User Interface

To have a user-friendly interface, we created a Streamlit web app that would allow the users to input their budget, choose an objective, and get the optimal investment amounts according to the model, as well as expected buys/views (based on the chosen objective). This image below is for a basic model only with one budgetary constraint. Additional constraints can be added based on the user’s needs.

Impact and Conclusion

A Deloitte 2023 CMO Survey said that marketing makes up 13.6% of a company's annual expenditure. In the age of data-driven strategy, our modeling framework can help executives and decision-makers efficiently allocate marketing capital given a company's constraints and goals. Our model achieved up to a 14% increase in views for the same budget, and up to a 23% increase in buys, again given the same budget, when compared to the baseline. We also explore how a user could customize these models to best fit their needs using constraints and extensions, such as a time or a region index or a multi/weighted objective function. In conclusion, given the correct data, optimization is a useful framework that can be effectively applied to create business efficiencies in the marketing field.