Could Mac Users Use Some Help Installing Firefox?

Ken Kovash

7 responses

We recently dug into the user’s experience when downloading Firefox, answering questions such as, “of people clicking the Firefox download button at mozilla.com, how many receive the full file and how many complete the installation process?”

With the latest Funnelcake shipment (January 13th), we wanted to take these insights a bit further, and breakdown that question above in several dimensions – by OS, by browser, and by geographical region.  In this post, we’ll focus on the first dimension, i.e., how does the experience of users differ based on operating system?

To start, let’s look at the distribution of users who clicked the download button (on the mozilla.com homepage or main Firefox product page).

There’s nothing too surprising to note here.  This distribution should come close to mirroring the overall population of Firefox users and the population of internet users more generally.

Next, let’s see exactly what happened at the moment of clicking the download button.  You’ll recall from our previous post that we asked, “for every 100 people who initiate the download process, where exactly do they end up just a few moments later?”, and the answer was visualized through the diagram below:

(Note: while the percentages above highlight November’s Funnelcake edition, they remained almost identical in January).

Moving onto the punchline, below is the breakdown of these numbers by OS:

A couple key things to note:

  • There seems to be a substantial difference between Mac and Windows users.  While both groups are able to successfully receive the file at roughly similar rates, the experience of users diverges once they move from downloading to completing the installation.
  • We’re not entirely sure what’s going on with the Linux numbers.  Much more digging needs to be done.

Update: First, we should explain how we measure whether or not a user installs Firefox.  We count an installation as successful when a user visits the Firefox firstrun page.  Given this fact and considering Alix’s point below, it’s clear that Mac installations will be undercounted in this analysis, as Mac users have to manually launch Firefox for the first time in order to be counted as an install, while Windows users automatically hit the firstrun page.

7 responses

  1. Majken “Lucy” Connor wrote on :

    With linux I’d wager that it’s new linux users who know they want Firefox but don’t know how to install apps on linux yet. They come to us, download the file, then when they get help installing Firefox they’re shown how to do it from the package manager.

    Would be interesting to see the linux numbers broken down by distro.

  2. Alix Franquet wrote on :

    The experience on Mac is different than on Windows. To install on Windows you need to open the set up wizard and go through the different steps, the last step has a checked box that says “Launch Firefox”. On the Mac, once the download is complete, one has to open it and drag and drop the Firefox icon into the application folder, but Firefox does not launch automatically. So we can’t tell in the Mac case if people didn’t install or if they installed successfully but never launched Firefox.

  3. Donnie Berkholz wrote on :

    “This distribution should come close to mirroring the overall population of Firefox users and the population of internet users more generally.”

    I would expect the Linux numbers wildly underestimate Firefox users on Linux. Perhaps the best estimate of this would be the Firefox upgrade page, which will show up to most Linux users regardless of where they got their Firefox.

    I also agree with Lucy’s comment regarding the Linux users who do download it here.

  4. Corey Burger wrote on :

    Like the download stats of anything that is packaged by a Linux distro, the numbers lie horribly, because the vast majority of Linux users receive their copy of Firefox|OpenOffice.org|etc. from the distro maker, not Mozilla itself.

  5. Gervase Markham wrote on :

    Perhaps before people download on Linux we should be suggesting that they get a package from their distro? After all, installing the .tar.gzs we ship is not exactly trivial.

    But then why does anyone download on Linux? Who doesn’t ship Firefox? Could it be that people are missing the in-product update mechanisms? Or are distros slow in providing new versions such that people feel they need to install them over the top?

    Or perhaps we should be providing packages for major distros ourselves.

  6. Mark wrote on :

    Alix noted that “…we can’t tell in the Mac case if people didn’t install or if they installed successfully but never launched Firefox.”
    I suppose at one level we would really like to know this, on another, it does have the same net effect – non-usage.

    I’m curious, is there a reason that we don’t use an installer on the Mac? Perhaps there’s no technical reason to do so, but it would allow us to do two things: 1) give a controlled, pleasant feeling install (so we’re hand-holding and informing/introducing) and 2) start the browser after the install.
    I think this would help our adoption rate by improving the overall first run experience and by making sure that the first run actually happens.

  7. Jesse Ruderman wrote on :

    I think the dmg distribution makes it unnecessarily hard to install Mac Firefox, at least from Safari. Many newer apps are distributed using simple zip files. Can you throw together a “Firefox.app in a zip” download for the next point release and do some A/B testing?