In the following post, we offer the data and the code to replicate the figures regarding CMP adoption included in the paper “Consent Management Platforms: Growth and Users’ Interactions over Time“, currently under revision.
In summary, the paper shows how Consent Management Platforms have become popular in the last 9 years, and are nowadays used by 42% of websites based in GDPR countries. The paper also analyses user behavior with Privacy Banners, showing how the banner design and the presented options impact the user’s choice.
Data and code are available for download at the following address: https://mplanestore.polito.it:5001/sharing/hn3CPuaNu
CMP popularity evolution
The datasets
The repository includes three datasets:
- HTTP-9Years summary: it includes the HTTP-9Years dataset in a summarized form. For each website, it reports uniquely the number of CMPs and Trackers
- HTTP-9Years detailed: it includes the HTTP9-Years dataset in a detailed form, as it indicates which CMP and Tracker is present for each website
- HTTP-Dec2023: the dataset for the last snapshot, containing the top-1M websites worldwide
The following table summarizes the salient features of the dataset:
The code
We offer a Jupyter Notebook to reproduce the plots in the paper. You need Python3, Jupyter Lab, and the libraries pandas and fastplot.
Running the Notebook, you will obtain the plots included in the paper, such as the following one, showing the CMP adoption rate in various countries of Europe.
User behaviour
The dataset
The repository contains a minimal, private dataset which logs users’ clicks to Privacy Banners. It contains the following features:
siteId: an integer represeting the website where the click happened;
createdAtDate: when the click has happened;
country: the country of the user who takes the choice;
region: the geographical macro-region of the user;
os: the operating system of the user;
browser: the browser employed by the user;
position: the position of the banner on the user’s screen;
banner: the banner version with which the user
blurred: whether the background was blurred when the banner appeared;
status: the choice of the user.
The code
Again, we offer a Jupyter Notebook to reproduce the results shown in the paper. For instance, the following plot show the percentage of users rejecting the use of cookies, divided by the country they connect from.