Skip to content

Brave’s Capture of Usage Data

Aubrey Keus edited this page May 4, 2021 · 20 revisions

Note: This document is a complement to the referral system described at https://github.com/brave/brave-browser/wiki/Brave%E2%80%99s-Use-of-Referral-Codes

Why does Brave capture usage data?

Usage data is used to guide business decisions, inform product development and guide support resource allocation.

What are usage requests?

A usage ping is a HTTP GET request, with data encoded in query parameters, used to count active users. The encoded request is sent to laptop-updates.brave.com either when the browser starts for the first time in a day, or after midnight if the browser is left open. It contains no personally identifying information.

Note: All fields are described in the table below with example values.

The request is encoded with the following rules:

  • The platform, channel, and version strings are retrieved from browser state
  • The daily flag is set to true
  • The weekly flag is set to true if the browser is used for the first time in a calendar week, starting on Monday
  • The monthly flag is set to true if the browser is used for the first time in a calendar month, starting on the 1st
  • The first flag is set to true if this is the first day the browser was installed
  • The woi (week of installation), ref (referral code) and dtoi (date of installation) are retrieved from browser state
  • The adsEnabled flag, set to true if ads are enabled in the browser.
  • The arch field, which has the CPU architecture Brave was compiled for. This is only sent if ads are enabled in the browser.

The request is routed to a CDN before it is forwarded to Brave for capture. The CDN performs the following:

  • Infers country code from IP address (countries with a small number of users < 1000 are excluded)
  • Infers region code (https://en.wikipedia.org/wiki/ISO_3166-2) from IP address
  • Removes IP address from data sent to Brave

The CDN modified request is sent to Brave for data capture. Brave performs the following at capture:

  • The current date (year_month_day) is stored

Once a day Brave aggregates the captured data, storing it for analysis and reporting.

What information is contained in a usage request?

Field Description Values Source Example
year_month_day Date of request capture YYYY-MM-DD encoded date Brave 2020-11-19
platform Platform identifier Winx64-bc, winia32-bc, linux-bc, osx-bc, android-bc, ios Browser winx64-bc
channel Channel identifier Nightly, dev, developer, beta, release Browser release
version Version number x.x.x semver formatted string Browser 1.2.3
daily Daily flag true or false Browser true
weekly Weekly flag true or false Browser true
monthly Monthly flag true or false Browser false
first First day of installation flag true or false Browser false
woi Week of installation - date of the first Monday before the installation date YYYY-MM-DD formatted date Browser 2020-11-16
ref Referral code (now limited to a small set of 50 to 100 referrers) ABC123 formatted string or the value none Browser ABC123
country_code Country Code 2 digit string containing the country code CDN US
region State / Province Code 2 digit string containing a sub-national region code CDN CA
dtoi Date of installation - date of the browser installation (held in browser state for 14 days then removed) YYYY-MM-DD formatted date Browser 2020-11-18
adsEnabled Ads status flag true or false Browser true
arch CPU architecture Brave is compiled for X86, x86_64, ia64, arm64, ppc64 Browser x86_64

What aggregated data is built from the usage requests?

Metric Description Rule
DAU Daily active users Count of requests for a year_month_day
DNU Daily new users Count of requests for a year_month_day where daily is true
DRU Daily returning users Count of requests for a year_month_day where daily is false
WAU Weekly active users Count of requests for a year_month_day where weekly is true
WNU Weekly new users Count of requests for a year_month_day where weekly is true and first is true
WRU Weekly returning users Count of requests for a year_month_day where weekly is false
MAU Monthly active users Count of requests for a year_month_day where monthly is true
MNU Monthly new users Count of requests for a year_month_day where monthly is true and first is true
MRU Monthly returning users Count of requests for a year_month_day where monthly is true and first is false
Daily retention Percentage of installations that use the browser a specific number of days after installation (capped at 14 days) Count of requests for a year_month_day where dtoi is X, divided by the number of requests on X in which first is true
Weekly retention Percentage of installations that use the browser at least one time in a calendar week compared to the number installed on a target week Count of requests for year_month_day where woi is X, divided by the number of requests in the week starting on X in which first is true

Privacy Considerations

The Brave usage counting system is specifically designed to not rely on personally identifying information. The system builds usage counts by aggregating the usage requests sent over the course of a single calendar day.

Clone this wiki locally