Changes Between Firefox 2 and 3

Now that we’ve taken a macro view of a few uninstall survey trends, we can dig into some of the details. The following two charts summarize the response to the second multiple choice question: “Why did you uninstall Firefox? (select all that apply).”

The initial results are encouraging. Fewer users uninstalled due to printing, security, usability, missing features, and web page rendering. Furthermore, more users were planning to return to Firefox! Perhaps the biggest surprise is the 2.6 percent increase in users uninstalling due to performance. We do not know whether this increase is due to (perceived) slower performance, to more crashes, to installation difficulties, or to a change in expectations.

While the response trends are generally positive, the survey itself has a number of areas for improvement. I have identified three areas where the uninstall survey falls short:

  • The question wording is unclear (performance, security, etc.)
  • The responses are clustered around 12% and don’t provide a clear area of focus
  • The responses are broken down by neither geography nor localization

Perhaps most troubling is the low response rate, depicted in the chart below.

Of 113,000 visitors to the uninstall survey in March, fewer than 8,000 left a multiple choice response. And, of those respondents, only 826 left a comment. I expect that the low response rate is due to the fact that that the survey is not localized. Nearly 90 percent of visitors to the uninstall survey live in non-English speaking countries.

In the next post I will suggest new questions for the uninstall survey and propose survey tool solutions that may help Mozilla address these areas of concern. Please leave your thoughts and suggestions for uninstall survey improvements.

10 responses

  1. Jesse Ruderman wrote on :

    The word “performance” might be too geeky. How about “Browser speed”?

  2. Gru wrote on :

    Actually Fx3 does seem slower/laggier in several areas. Most annoying is when it continuously accesses the hard disk (e.g. Fx2 was usable as soon as the main window came up. With Fx3, even after the main window has appeared, I have to wait for the disk thrashing to stop before it becomes responsive.)

    Also, if you try Fx3 on a really slow machine (e.g. an old PII), it will feel really slower than Fx2 … I wouldn’t recommend Fx3 for people with old computers …

  3. Simon Geard wrote on :

    Might the low response rate not be just that people uninstalling a program have no particular interest in filling out a survey? Seems a more likely explanation than lack of localisation…

  4. Simon Geard wrote on :

    Oh, and the comments form on this blog is awful. With no text captions next to all the fields (just small and vague icons), I had to use firebug to work out what each one was meant to be. Not really a very good look…

  5. Sean Hogan wrote on :

    RE: Performance
    In my experience FF2 is more predictable:
    – It uses a lot of memory, slows down and eventually crashes.
    – Or I restart it because it gets too slow.
    – Occasionally it crashes (or stalls) on a certain page, but that is fairly repeatable.

    FF3 doesn’t chew through as much memory nor get as slow, but it crashes more frequently.
    The worst thing is that after a lot of page views it starts to exhibit unreliable behavior, including
    – page refreshes stop working, or are unreliable
    – navigating to a new page fails, or it might require pasting the link into a new tab

  6. Simon Geard wrote on :

    Sean – I’ve found the opposite myself, with regard to crashing. FF3 crashes very rarely, much less so than FF2…

  7. palmiche wrote on :

    A month ago I posted bug 449919 about FF3 perf compared to FF2. My guess is something got messed up in the rendering. FF3 is more stable, etc. but is so sloooooow. No matter how many JS optimizations you get they’re worth nothing if an HTML-only page slows down your browser.

  8. Gru wrote on :

    Crashes can be due to binary extensions, too. I am very satisfied with Fx3’s stability. But then I usually don’t keep the browser open for 4+ hours …

  9. Hescott wrote on :

    Hey, I think you should absolutely add “Localisation” to the reasons for uninstalling the browser — combined with what locale the user was on, this would show when localisations are in bad shape.

  10. Harry wrote on :

    Is the response rate really troubling? It looks a bit like an instance of the 1:10:100 rule.

    A study of Yahoo groups participation (in 2006) gave a roughly similar pattern: