My favorites | Sign in
rlz
Project Home Downloads Wiki Issues Source
Search
for
HowToReadAnRlzString  
How to interpret the parts of an RLZ parameter string
Updated Jun 25, 2010 by gwilson%chromium.org@gtempaccount.com

An RLZ URL parameter is a string consisting of many parts:

  • rlz encoding is at version 1 as of Sept 2007, but this will be incremented each time the encoding scheme changes.
  • application T=Toolbar, B=Firefox Toolbar, D=Desktop (authoritative list in the source code) With one digit for major version.
  • brand code identifies distribution channel, may be a partner or internal marketing. This correlates to how the user got the software (ie. they downloaded it by itself vs. it came pre-installed on their new computer vs. it came with a partner's software).
  • cannibal tells if the library has evidence that the user was a user prior to installing the software. It's from the term "cannibalization" which refers to the fact that users who get client products through distribution were already loyal customers, thus impacting the actual incremental value of the channel.
  • language is the two- (en) or five-character (zh-CN) language code of the application. Valid values depend on the specific app.
  • install cohort is the country and week of the user's installation event. Country is determined by the server, using IP address. Week is measured as number of weeks since Feb 3, 2003. It's used to measure attrition rate and is used in ROI and accounting analyses.
  • first search cohort is just like install cohort, but for first search event. They're both called "life cycle events".

Other specifics of RLZ parameter strings:

  • rlz encoding, application, brand code, language, and install cohort are the required fields, and are the minimum that will appear in an RLZ string.
    • Underscores generally indicate a missing value. An underscore where the cannibal signal should be indicates that this is not a cannibalization.
    • first search cohort may not appear. If only one 'cohort' value appears in a string, it must be the install cohort
  • The brand code can be any four alphabetic characters, all upper-case. There is not exhaustive enumeration, as this can be any four-alphabetic-character value you want.
  • The only rlz encoding value at this time is "1".

For example, if the string reads:

rlz=1T4AAAA_enUS236

This means this installation was not a cannibalization and is not (yet) part of a first-search cohort.

In a very small number of cases, if network or registry key interactions fail, the install cohort will be replaced by underscores, as so:

rlz=1T4AAAA_en_____US239

These instances are rare, however.

Comment by jsmelley, Jun 3, 2010

Is there a listing of the known "brand code's"? Also, how are null values represented and what would the absence of a value indicate? If a value is null is its place filled with an underscore("")? I have seen some RLZ values that only have one "US###" value and I am trying to determine if it is the "install cohort" or the "first search cohort" Is there a way to determine this? Lastly, is there a listing of known RLZ encoding values and their meanings?

Comment by jdr0...@yahoo.com, Jun 11, 2010

Agree with jsmelley, the Brand Code field would provide valuable information, but of course, it's still a cryptic field.

Comment by gwilson%chromium.org@gtempaccount.com, Jun 18, 2010

jsmelley, see the latest edits to the wiki above (hopefully that answers your questions)

Regarding brand codes, these can be any four upper case alphabetic letters. It's just an identifier, and there's no pattern or ordering to the letters.

For example, the brand code "GGLS" -> downloading Google Chrome from www.google.com/chrome. But the individual letters and ordering don't mean anything.


Sign in to add a comment
Powered by Google Project Hosting