1689106 - The attribution is not added fore new users if the Firefox stub installer is downloaded directly from the Mozilla "download/thanks/" page

Status: NEW

(www.mozilla.org :: General, defect)

Description

[Cosmin Muntean [:cmuntean], Ecosystem QA]

Attachment: attribution not added.gif

[Affected versions]:

• Firefox Realease 85.0;

[Affected Platforms]:

• Windows 10 x64;
• Windows 8.1 x32;
• Windows 7 x64;

[Prerequisites]:

• The Firefox browser is uninstalled, all the registries and all the appdata are removed.

[Steps to reproduce]:

1. Open the Chrome or Edge browser.
2. Navigate diretly to https://www.mozilla.org/en-CA/firefox/download/thanks/ page.
3. Wait for the stub installer to be downloaded.
4. Install the Firefox browser.
5. Open the Firefox browser and navigate to "about:telemetry#environment-data-tab_settings" page.
6. Observe the "attribution.ua" property.

[Expected results]

• The "attribution.ua" property exists and has the "chrome" or "edge" value.

[Acutal results]:

• The "attribution.ua" property dosen't exists.

[Notes]:

• The issue is not reproducible if the Firefox stub installer is downloaded from https://www.mozilla.org/en-US/firefox/new/ page.
• The issue is not reproducible for Firefox Beta, Developer Edition and Nightly builds downloaded from https://www.mozilla.org/en-US/firefox/channel/desktop/.
• Attached a screen recording with the issue.

Comment 1

[Alex Gibson [:agibson]]

FWIW, this is intentional by design, at least when the original stub attribution feature was implemented. We don't block downloading Firefox whilst waiting on a stub attribution auth request. That request has to happen before the user clicks download on a button that points to /thanks/.

Comment 2

[Cosmin Muntean [:cmuntean], Ecosystem QA]

Thanks Alex for your input.
We were inclined to log this behaviour because there is an experiment that targets the new users who switch from Chrome to Firefox. Considering this, there is a chance that if new users receive or download the Firefox stub installer from the /thanks/ page, they will not have any attribution set, therefore they would not be eligible for the experiment. The experiment I am talking about is “Multi-Stage Onboarding : Custom Chrome strings”.

@emily could you share your thoughts on this? Would it be possible for this to lead to an improvement in our Chrome switcher detection?
Also, feel free to add anyone else who could weigh in on this as well.

Comment 3

[Emily McMinn :emcminn]

Hi Cosmin, I'm unfortunately not the right person to answer here, I've only touched attribution in the context of this experiment. I'll pass the needinfo on to @Mardak who has worked on it in more depth :)