Marketing Mix Modeling
By Kevin Eddy
For this project I'm using the dataset found in the following link: https://www.kaggle.com/datasets/orosas/marketing-mix-dataset
Given the proliferation of marketing channels, it is increasingly important for marketing teams to understand the impact of different types of advertising on sales. This project uses data from Facebook, TikTok, and AdWords to help the marketing team to:
- Identify the most effective marketing activities for driving sales.
- Optimize their marketing budgets by allocating more money to the most effective activities.
- Measure the impact of marketing campaigns on sales.
- Forecast future sales.
The Goal
In the end, we want to be able to answer questions such as: How much of the $ 15,904.11 sales in the week ending on 2021โ10โ10 was generated by Meta advertising? And how much by TikTok and AdWords? And what is the baseline, i.e. the number of sales we would have had without any advertising?
If our model can accurately predict sales, we can use it to calculate return on investment (ROI) and optimize spending, which are the ultimate goals of most companies.
Additive models are best suited for MMM because they allow us to easily decompose sales. This means that we can identify the specific impact of each marketing activity on sales. Other models, such as random forests, gradient boosting, and neural networks, do not allow for this decomposition.
For the first half of this analysis we will use a linear regression model because it is the simplest representative of additive models.
In the second half we will improve our model to include advertising adstock which will allow us to imcorporate the effect of diminishing returns.