Why People Don’t Install Firefox – Part II

As we described a few days ago, we recently shipped a modified version of Firefox with a different installer – one allowing users to leave feedback upon canceling.  In Part II of this series, we want to outline exactly what users saw and share some high-level findings.  An examination of users’ written feedback – and a discussion around associated insights, implications and next steps – will soon be coming in Part III.

What are all the steps of the Firefox installer (Windows)?  Where exactly can a user hit “cancel”? (Please click below for the full image)


When a user clicked “cancel”, what exactly did he/she see? (Please note that clicking on any sub-category, such as “Taking too long”, opens up a free form feedback box for the user to write their comments)





What were the high-level results from this feedback form?

Below is a screenshot from Kampyle’s dashboard (click for full image).  Kampyle, a start-up making huge strides within the analytics world, helped Mozilla with this initiative by hosting the feedback form you see above.  You’ll notice a few high-level pieces of info:

  • What was the average sentiment/feeling of people upon canceling?
  • How many total people left feedback?  (Answer: about 5,000 people hit the feedback form and about 330 left feedback)
  • What were the most common “categories” of feedback left by users (i.e., types of problems cited)?
  • What steps within the Firefox installer saw the most users abandoning?  And what was the feedback breakdown for each of these steps?


In our next post, we’ll dive into the free form feedback left by users (where we can more clearly identify a couple critical trouble spots) and discuss what it all means for possible product and/or marketing changes in the future.

Stay tuned!

15 responses

  1. Justin Scott (fligtar) wrote on :

    These cliffhangers are killing me!

    (and so is this comment box. really cryptic UI)

  2. Justin Dolske wrote on :

    “about 5,000 people hit the feedback form and about 330 left feedback”

    I wonder if more feedback would come with a simpler feedback window… It’s seems rather busy and complex.

  3. Kurt wrote on :

    Yeah, it is very busy!

  4. david dahl wrote on :

    I love my “apt-get install firefox”.

    Perhaps if the first “wizard” window had a button saying “just install Firefox the way most of the 190 million Windows users install it”, we would have fewer aborted installs? Do we track what “most users” select in this journey of an install process?

    Just my 2 cents.


  5. Otto wrote on :

    The screenshots for step 4 and 5 are identical.

    I was wondering why step 4 didn’t see so many people give up, so it would be nice to be sure what step 4 was.

  6. JG wrote on :

    I think you’d be better off with just a simple freeform text box and doing a qualitative analysis on the answers rather than quantative. e.g. “We noticed you canceled your installation. We’d really appreciate you taking a second to quickly tell us why. Thanks!” This puts the work on you, which is where you want it when you’re trying to gather user feedback. Put the work on the user, and they won’t do it.

    Your average user can’t be troubled to click more than twice, but give them a text box and they’ll type til they run out of space…

  7. JG wrote on :

    Oh, and this seems obvious, but are you measuring what page the users are canceling on? That will likely give you more useful information than the actual user feedback. Actions speak louder than words, as they say. And knowing your funnel (what % of users you lose at each step) is the most important thing in user acquisition analysis.

  8. alex.r. wrote on :

    The feedback UI is way too complex. I don’t know what percentage of “I’m confused” were actually referring to the feedback program instead of the installation program.

    Step 4 and 5 look exactly the same? Why the discrepancy in the number of cancellation between the two?

    The large amount of cancellation on step 2 might be strictly related to the fact that it’s the first thing that users actually read. Step 1 is pretty useless as far as I’m concerned.

  9. kkovash wrote on :

    Otto and Alex — sorry about the image showing steps #4 and #5 as being identical. The image has been adjusted and is now correct.


  10. Robert Strong wrote on :

    The high cancellation rates at step 2 is very likely due to the firefox.exe process continuing to run after the user exited. The installer doesn’t forcibly close Firefox due to potential dataloss.

    The high cancellation rates at step 3 is very likely due to the user not having admin rights and the installer requiring the user to select a directory where the user can write to.

    We have been talking about lessening the number of installer pages in the near future.

  11. Anonymous wrote on :

    I just read all 6 steps of the install process. It looks 5-6 steps too long.

    Step 1 looks completely worthless, as it doesn’t ask any questions. Step 6, similarly, seems unnecessary. Steps 3-5 should not occur in a normal install.

    Suggested replacement: when you run the installer, show *one* step similar to step 2, with a big friendly “Install Firefox” button, and a smaller twisty for “Edit installation settings” or similar. (Don’t use “custom” or “customize” here, because Firefox has various campaigns to encourage people to customize their browser, and you don’t want people thinking they need to choose “custom” unnecessarily.) If the user hits “Install Firefox”, the install proceeds with no further interaction. If the user clicks “edit installation settings”, it expands to show the checkboxes for icons and default browser, the destination folder, and the start menu folder, as well as a “use default installation settings” option. After changing those, the user can then hit the same “Install Firefox” button.

    This way, you have *one* dialog box, *one* click to install with the defaults, and easy accessibility to all the custom installation options at once.

  12. Dan wrote on :

    The percentage of people responding does seem low, and I’d say that’s because the feedback form needs a better design, more simplicity.

    The reasons that Robert Strong listed above need to be included so that you can tell if that’s why they canceled or not.

    Answers (to the question “why did you cancel?”) should also include:

    Firefox is already installed, oops.

    Takes too long to install.

  13. Bob Arnson wrote on :

    Another datapoint to capture is the amount of customization when users take that path. For example, I always choose a custom install because I always deselect the desktop and quick launch shortcuts (because I only ever use pinned root start menu items [or urls or hardware buttons or…]). But I never change the installation directory or start menu folder. Does anybody change the start menu folder?

  14. name wrote on :

    Come on, it’s just stupid saying that people don’t use Fx because some installer steps are unnecessary. I despise installers that don’t let me choose the start menu folder where the shortcut will be created.

    Also, instead of the installer, you should work a bit on this comment form.

  15. kkovash wrote on :

    Hi there,

    sorry about the comment form UI. I’m working on getting it fixed/improved.