Recent eye tracking studies on the web confirm your assumption, the ‘heat’ area is the upper-left corner :
http://www.websiteoptimization.com/speed/tweak/clickstream/

I like the boldness and simplicity of the design. The large Firefox logo provides instant impact and is a good reinforcement of the branding for potential users who may have been exposed to it elsewhere. The “Download Firefox today!” headline makes it clear what you’re here for upfront.

I also really like the attention-grabbing “FREE” box. It’s better than “Free Download”, which has unfortunately become a euphemism for “You can download it for free but you’ll have to pay later”.

The text, however, is terrible. Comparing to IE alienates anybody who doesn’t use IE, which includes Mac users (IE is an ex-browser there) and those who worship at the alter of Tux.

I’ve always thought that promoting safety and security is a really bad idea, as it just invites ridicule when there’s a security flaw. It’s also very negative: even just saying that Firefox is secure has the implicit message that IE isn’t. With Vista’s Protected Mode (which Firefox doesn’t use), I’m not sure that’s such a valid argument anymore.

There’s also “And no more pop-ups!” A pop-up? What’s a pop-up? IE has had a pop-up blocker for almost three years. Dead. Horse. Flogging. Stop.

Finally, the links to things like the release notes, other versions and system requirements are too far away from the download button. I use the other versions link quite frequently, mostly because the locale detection on the page is pretty sucky (granted, I use it because I know it’s there but let’s not make it harder for the novice user).

The page also needs links to features pages etc., for the undecideds who want more info before downloading (right now, the page is a bit of a dead end).

Test Version #2:

It has a screenshot, which is good. Bear in mind that some people arriving at this page won’t know what a browser is, so a screenshot is a good way to communicate this.

The feature bullet points are nice in a why-should-I-download-this sort of way but they do make the page look a bit busy.

“Rest easy knowing we block… viruses and spyware.” No, we don’t. Saying that is downright irresponsible.

Moving beyond the two sample designs specifically, I think linking to pages with details about features is useful and informative but it does lead people further away from the download page. Perhaps some sort of DHTML overlay system could be utilised to provide more details (maybe even demo movies) without taking potential users away from the download page.

Overall, the download page has to do two things. One is to promote the browser. The other is to reassure people that it’s safe to download. Users have to be assured that it’s really free, can import their settings, will coexist with their current software and won’t screw up their computer.

So many downloads today are really ploys to get users to commit to something else that the idea of a download with no ulterior motives is alien to many people. It’s hard to convince people of the altruism of the Mozilla project. Tragically, that’s the attitude that we have to fight against.

And that’s why I believe in Mozilla.

1) The current version (for left to right readers) has an advantage, I think, in that the download button is significantly closer to the top left corner of the page. I may be wrong on this (I bet Alex Faaborg would know), but I would wager that people, faced with an otherwise unremarkable page, tend to look first at the upper-left quadrant of the page, then upper right, then lower left, then lower right. Putting the download button in the lower-right quadrant seems like possibly the least desirable thing, if my assumptions are correct.

2) I find the big dark-blue “Free!” block in Test Version #1 to be significantly more eye-catching than the green download button. Perhaps the pale green isn’t the optimal choice for button colour? With the rest of the colours on the page, it sort of fades into the background and isn’t that obvious. Maybe looking at the page designs with an eye towards contrast and composition (as if they were a black and white photographs) could help ensure that we’re catching and moving the reader’s eye from the primary to secondary to tertiary points of interest as we intend?

3) I do like the “below the fold” design of the Test version #2 with the two columns with icons. Ensuring those icons match up with the icons/metaphors we use in the browser could be a great way to really reinforce those concepts in the user’s mind.

4) I don’t think we should talk about competitors or competing directly like we do with the “big blue e” thing (trademark usage issues aside). I could go on and on about this but I think there’s a fundamental conceptual line here that we probably don’t want to cross (I could be wrong!)

5) Saying that “Firefox is stuffed” is going to simply add fuel to the “Firefox is bloatware!” meme that is growing across the internet. We should probably try to avoid that.

6) I’m not sure how much input you’re looking for regarding the copywriting in these. I have some feedback if you’re looking for nitty-gritty wordsmithing-style comments.

7) Generally the test pages are very pale/pastel outside of the Firefox logo. Punching up the colour couldn’t hurt.

One really key point that I forgot to mention in my original post is that because these are just test pages, we’re only serving them up to a relatively small audience (~10,000 page views each). Therefore, given that our resources are limited, we’re only running this test for the en-US version of the site.

Once we feel like we have a version that works well (may take several tests to do this) we’ll begin working through all the localization issues. We definitely hadn’t forgotten that part, though, don’t worry!

Also, I should stress once again that this is going to be a multi-step process of testing and testing until we get it right. If some of the language or imagery seems different than usual it’s because we’re using this opportunity to try some new things to see if they have any impact. We’ll refine as we go and hopefully will wind up with something that increases the download rate from this page. I’ll post the initial results in a few weeks…stay tuned.

Thanks!
John

Definitely. And I believe a screenshot is a must as it is not obvious in the other versions (current and test #1) what a web browser is about or what “more to the web than the big blue e” means: is Firefox a plugin for IE then?

Not sure how hard it could be in legal terms to feature a variety of well-known web sites in the screenshots like Google Homepage, My Yahoo, Netvibes, Pageflakes. People will more likely relate to an application “for” actual content they are interested in. At the same time we must be careful not to alienate people who prefers a competing service.Or perhaps create a mockup web site that looks more real an interesting than “Getting started with Firefox”. It could be a fake news site with snippets from here and there.

Another enhancement would be to detect the browser the visitor is using and present the features that may be relevant for him. IE7 users already have tabs, and popup blocking, so either we say it blocks more pop ups than IE 7 or move it to a lower position in the page.

I’d also mention themes among the add-ons and even better would display a few if space allows. Spell checking should also be mentioned but not as “it includes a spell checker”, but something like: “no more messy emails/posts/typos”. Just like with “be smart” we don’t want to tell users their spelling sucks.

And as Marcia mentions, test#2 needs a human touch, but in all I prefer it than the other two.

Mainly for these reasons :
1/ The XP userbase is rapidly declining in favour of Vista, this is mechanical since XP is no longer sold with new computers
2/ We need to show our Vista compatibility and outline the fact that we also target IE7 users,
3/ In a multilingual web presence you create web content with the longest life-expectancy, re-creating screenshots with graphic design works when Vista is majoritary is costly in terms of money and time

The perfect solution would of course be to detect the OS via the user agent and send the user’s OS version of the screenshot.