A research-oriented top sites ranking hardened against manipulation

Researchers in web security or Internet measurements often use rankings of popular websites (e.g. the Alexa ranking). However, in our paper we showed that these rankings disagree on which domains are most popular, can change significantly on a daily basis and can be manipulated (by malicious actors).

As the research community still benefits from regularly updated lists of popular domains, we provide Tranco, a new ranking that improves upon the shortcomings of current lists. We also emphasize the reproducibility of these rankings and the studies using them by providing permanent citable references.

Access the Tranco ranking

We advise to use the latest standard Tranco list: one million domains obtained by combining all four rankings averaged over the past 30 days.

You can directly retrieve the latest Tranco list at this permanent URL, e.g. for use in daily crawls. You can retrieve the permanent ID of this list separately to refer to it later on.

The file uses the same format as the Alexa and Cisco Umbrella lists. A daily update to the list is made available by 0:00 UTC; the Last-Modified header provides an exact timestamp.

You can install the tranco Python package to directly access the list from within your code.

You can retrieve past versions of the standard Tranco list.

You can customize your list to have specific properties.

Configure a custom list

Attribution: We use the lists from four providers: Alexa, Cisco Umbrella (available free of charge), Majestic (available under a Creative Commons Attribution 3.0 Unported License) and Quantcast. Tranco is not affiliated with any of these providers.

News

We presented a long-term evaluation of Tranco at the 12th USENIX Workshop on Cyber Security Experimentation and Test.
We posted a summary of our findings on the RIPE Labs blog.
We have enabled filters on Google Safe Browsing (to remove malicious websites) and the Chrome User Experience Report (to keep only sites regularly visited in Chrome).

Usage of Tranco

Over 20 academic publications have referenced and/or used Tranco.
Mozilla runs a regular scan on Tranco to find websites using TLS 1.0 or 1.1.
The Norwegian public broadcaster NRK used Tranco to study cookies on Norwegian websites.
The NoLeaks project analyzes tracking on EU domains in Tranco.
Scott Helme scans Tranco to assess the state of security on the web.

If you're using Tranco, let us know!

About this service

Researchers often use rankings of popular websites when measuring security practices, evaluating defenses, analyzing ecosystems or conducting Internet-wide observations. However, little is known about the data collection and processing methodologies of these rankings, even though this affects the validity of conclusions based upon the results obtained using these rankings.

We uncovered how both inherent properties and vulnerabilities to adversarial manipulation of these rankings may affect the conclusions of security studies. In addition, we found that researchers overwhelmingly lack any comment on when the list was downloaded, when the websites on the lists were visited and what proportion was actually reachable. This hampers reproducibility of these studies given the daily changes in list compositions and ranks.

Overall, our study revealed significant shortcomings of current rankings and calls for a more cautious approach from the research community. In order to make researchers more aware of the properties and effects of these popularity rankings and to help in ensuring that studies using these rankings are reproducible, we provide this service where researchers can obtain rankings with properties that are more appropriate to their studies, and where the methodology and composition of these rankings is stored for future reference and reproduction.

We propose a ranking that is suitable for most research purposes, consisting of data from all available rankings over a period of 30 days. We also allow researchers to customize rankings, applying multiple filters in order to best reflect their needs. We archive all generated rankings and supply a permanent link to a page with additional information on the methodology behind the ranking and the ability to download the exact same list as was used in the study. This improves the validity of research results through a better understanding and awareness of these rankings, as well as reproducibility by keeping a permanent reference to the rankings.

About us

Feel free to contact us regarding Tranco and our research.

Victor Le Pochat - [email protected] (main contact)
Tom Van Goethem - [email protected]
Samaneh Tajalizadehkhoob - [email protected]
Maciej Korczyński - [email protected]
Wouter Joosen - [email protected]

The source code for the Tranco ranking is available on GitHub. If you run into any problems with the ranking or the website, you can create an issue there.