Marketing Mix Modeling fight: Cassandra vs PyMC Marketing

Marketing Mix Modeling using Python - PyMC Marketing vs Cassandra a MMM software
Find the resources used in the video here:
Data request checklist: bit.ly/3Udaopi
MMM Dataset: bit.ly/44l5WcE
Colab (code): bit.ly/3QoroYE
Cassandra: www.cassandra.app/
PyMC Marketing Repo: www.pymc-marketing.io/en/stable/
Cassandra: cassandra.app
Here is the agenda of the video:
0:00 Intro
0:40 MMM Data Request Checklist
2:08 Installing pymc-marketing
4:15 MMM data cleaning - Python
6:15 Extract Seasonality function - Python
10:15 MMM feature selection - PyMC Marketing
12:55 Py-MC configuration & training
13:15 Cassandra MMM
13:48 MMM feature selection - Cassandra
14:31 Automatic EDA Analysis - Cassandra
15:03 Data iteration
16:00 MMM training - Cassandra
17:22 PyMC re-training after data iteration
18:02 Contribution Decomposition over time - PyMC Marketing
20:14 Diminishing returns - PyMC Marketing
22:13 MMM Prediction vs Actual sales - PyMC Marketing
25:22 ROAS probability distribution analysis - PyMC Marketing
26:00 Model selection - Cassandra
27:50 ROI analysis - Cassandra
29:00 AI MMM suggestions - Cassandra
30:30 Seasonality analysis - Cassandra
31:00 Contribution over time - Cassandra
33:00 Results comparison between PyMc Marketing & Cassandra
37:00 Predict on test data - PyMC Marketing
38:25 Budget allocator - PyMC vs Cassandra
42:00 Conclusions and evaluation grades

[@aaakmm1785]

What kind of data do you include in your MMM models? Do you include only paid media? Should organic and direct traffic (clicks) be included in the model? Is it a good idea to exclude from the model small channels with little traffic?

[@cassandra4533]

We have a complete tutorial series on building an MMM, starting from the dataset to be used.
Here's the video: th-cam.com/video/HE8qZSzPWfc/w-d-xo.html&ab_channel=Cassandra
In short: you should include all the variables that have an impact on your output KPI.

[@aaakmm1785]

@@cassandra4533 Thank you. In this example you only used paid channels, no direct or organic traffic. Was it on purpose or only an example?

[@cassandra4533]

@@aaakmm1785 Only an example

[@aaakmm1785]

Hi. Recently Google published "The MMM handbook" and "Modern Measurement Playbook". There they say "Lower-funnel MMM (more traditional MMM) typically undervalues brand media by only measuring the direct impact of media on sales." and as a solution they propose to estimate also what they call "nested MMM". Could you make a video showing how to do nested MMMs?

[@cassandra4533]

Hello there,
Google's refers mostly to traditional MMM that have been using a wrong methodology for adstock.
Traditional MMM uses geometric adstock effect of media campaigns.
This methodology does not take into account the delayed effect between the spend and the effect on sales generated.
This leads to overestimating Lower funnel campaigns rathern than upperfunnel's.
The easiest way to solve this problem is by leveraging weibull adstock, that takes into account that.
Nested MMMs are incredibly difficult to replicate, even more difficult to become actionable and useful only on handful of situations.
As soon as Meridian comes out, I'll do a video about it, and if it has a feature for nested MMMs I'll be happy to show it.