Case Study

Social Media Ads Forecasting

Social Media Ads Forecasting

Understanding the key driving forces behind an effective marketing strategy is essential to getting out the most out of advertising budget for business growth. Through this dataset, I have the opportunity to use data science to evaluate the performance of ads on social media platforms like Google, Facebook and Instagram and then propose a suitable ads spending strategy.

Here comes a business question: Which social ads platforms (Google, Facebook or Instagram) would you invest in if you are given an allocation of RM100k advertising budget?
And the business objective is to
1. Suggest an ad spending strategy
2. Forecast sales based on the recommended ad spending on social media platforms.

For the exploratory data analysis (EDA), a scatterplot with a regression line has been created using seaborn.regplot in Python to show the linear relationship between sales and each of the social media ads platform.

Besides having regression plots, I also used simple Linear Regression model to calculate the R-squared value of each social media ad to find out which ad is having the highest sales predictability.

FeatureR-Squared value

In addition to R-squared, I also used multiple Linear Regression model to calculate the coefficient value of each feature.


I also managed to come out with a proposed ads spending strategy after performing the data exploratory and data modelling.

  1. Due to highest predictability of sales from google ads, I will allocate 50% of the budget for Google advertising.
  2. Since Instagram is having -negative coefficient, I will only allocate 10% for Instagram advertising and the remaining 40% will be used for Facebook advertising.
  3. In terms of targeting, I will focus on large market size and urban area.
  4. With the mentioned budget allocations and targeting, the expected revenue is 65 Mil (assuming unit price is 5k).

If you want to know the steps on how I came out with my ad spending strategy, please do check out my data works here.

Published by Lee Hong

Leave a Reply

Your email address will not be published. Required fields are marked *