It’s actually not super difficult… if you have good data. The first step is to understand if you have marketing spend by channel for each channel at a weekly level. Check with the marketing leads.
To understand the cost of acquisition, you need to predict how many new users joined by spend. Run a regression model, with log independent variables, spend and hold for seasonality.
X Variables
Paid Search Spend log
Organic Search Spend log
Social Media Spend log
Offline – TV Spend log
Month or Quarter binary
Perhaps t for time trend (month X) continuous
Y:
N. of New Users