I've always seen a fairly steady Opera usage (>0.2%)with little to no rise or fall.
As a committed (and paid up!) Opera user I hope not!
Decent traffic site, older users: Opera: 0.2%, give or take, usually less.
Boring commercial / manufacturing site: usually 0%, sometimes an opera user or two visits in a week.
Standard spam type sites: 0%, almost never.
Technically oriented sites, around 2%, more or less the same as Safari. Firefox 40%, give or take.
I no longer support Opera unless it's a very high traffic site, or a technically oriented one, or unless I get a client who will pay for that support, which I've never had yet. I spend the amount of time their market share is worth, they had years to build it up and failed.
Personally, I'd rather my sites function decently in elinks, links, or w3m, and I'll usually spend at least some time to make sure they work for those users, since anyone using a console browser deserves extra credit, and a site that works well for text mode browsers works even better for search engine spiders.
This means some small display bugs remain in some cases, but the page more or less works, usually it's fine though, but I've found extremely simple css / html that completely breaks in Opera, even the latest release.
Sigh. I always thought Opera was awesome, but it certainly never caught on among the visitors to my sites, even though I even used Opera as my primary browser for a year or more.
They got many things right, long before Mozilla 6 even existed, let alone Firefox or IE 6.
I can't waste time doing security updates for friends and family and customers for a browser very few webmasters support. Its a continual hassle trying to keep IE, FireFox, and Opera all patched and "safe" to use.
Does anyone yet offer a "protected" mode browser for Windoze yet? Specifically, one that runs with in a limited security context and can not write to any folder other than those explicitly permitted. Various e-mule mods have this ability on Windows 2K and newer.
The default user agent string for Opera is set to Internet Explorer, so the figures in your log are probably not representitive of the true number of users visiting with Opera.
Yes it is, testing for 'opera' isn't rocket science, and my stat program has no problems with distinguishing between Opera and other browsers, all Opera user agent strings contain the string 'opera'. This is the argument always given when people don't want to admit how low the numbers really are.
Default Opera 8 useragent: Navigator User Agent: Mozilla/4.0 (compatible; MSIE 6.0; Windows NT 5.0; en) Opera 8.0
You think it's extremely difficult to distinguish between this string and the default IE string, but it isn't, I've seen very accurate opera numbers for years.
Both Opera and FF are virtually non-existent over all my sites. People here have been going on about the FF revolution but outside of this forum I don't know anyone that has even heard of it!
As with all these things, there are two main factors:
who your audience is
how well you cater to their needs
Example, I know at least one internet banking iste that would claim they only need to support IE because that is all their users ever use. In fact, they are giving away X% of the total banking market to sites that work with other browsers.
It's their choice, of course, subject to the inevitable class action from their shareholders who may feel that that choice directly affects their investment.
On a very techie site, I am seeing 6% Opera and 36% Firefox.
I've no idea if that is atypical of the world at large, or an indicator of how things may be changing.
It helps that the site is coded to strict standards, so it'll work with anything.
2by4, what stats program are you using?
On combined stats from five or six non-techie sites with about 200,000 unique visitors a month in total, I get about 8.5% Firefox and about 0.1% Opera.
Safari is surprisingly (to me anyway) high at 2.6%, with almost 6% of visitors using Macs. None of the sites have topics that might be 'mac leaning' - like publishing or graphics or such. Is this high or usual?
Analog, we use others, but the numbers are all the same, doesn't matter, urchin etc.
For some reason people think testing a string to see if it contains the substring 'opera' is difficult or somehow overwhelmingly complex, but let me assure you, it isn't. Anyone who can write a stat program can test for the presence of a substring value within a string.
to make it very simple, for those who don't know any programming at all, here's how you do it:
string = 'msie 6 opera 8';
is_opera = true
is_msie = true
and so on through all the options
Speaking for analog, they've correctly id'ed opera since at least version 3, probably before that, that's one of the most trivial thing a stat program can do.
Re Firefox stats, most sites I do have pretty much exactly the same firefox/netscape percentages, between 4 and 10%, across the board, except for technically oriented sites, where it pushes close to 50% at times, usually around 40%, same numbers reported by most users. Firefox has increased by about 4% on average over the last year, give or take a bit.
The real issue is whether or not your useragent total count has bots stripped out or not, for example urchin in default mode will give 50-60% MSIE because it also counts all the bot activity, which is pretty weak. But after doing the math the numbers always turn out to be about the same, across site, across topic, the numbers reported in the media are pretty accurate as a rule from what I can see, Opera .2% if you're lucky for standard sites, firefox/netscape 4-10% on average.
<added>By the way, the one browser that can avoid stat programs completely is Safari/Konqueror, where it has the option to give a user agent string that is an exact duplicate of the MSIE XP string, that's impossible to detect, so it's possible that safari numbers are actually slightly higher than might be reported.
As for users going in and changing their strings manually, the numbers of people who would know how to do that are so miniscule that I don't think they are statistically meaningful.
|As for users going in and changing their strings manually, the numbers of people who would know how to do that are so minuscule that I don't think they are statistically meaningful. |
I tend to agree with you about the statistically significant part. And yet I can tell you from the weekly analysis of user agents I do for my browser capabilities project that there are a lot of people who like to tinker with their user agent string. Sadly I often see profane things in the user agent string relating to the user's misconception of what a user agent string does.
From my perspective I don't see Opera fading at all. For years they represented less than one percent of my total visitors. As of last week Opera was at 1.1%.
I'd say it's fairly safe to say that the more technically oriented your site is, the higher your Opera numbers will be. Mine for example are about 10 times higher than normal, but so are my Linux users numbers. The less technical in subject matter your site is, the lower your opera user counts will be. Which is why 0.2% seems to be what Opera really averages on average sites, not tech types sites. The less sophisticated the market, the closer to 0 Opera gets, but oddly enough, Firefox has managed to grab some of that totally non-sophisticated market share, which impresses me, that's hard to do.
They've been much more focused on getting their browser embedded in smartphones and so on, than they are on the desktop marketplace. Interesting strategy and perhaps quite farsighted, even though it doesn't generate a lot of name recognition.
I for one hope thay continue to be a very healthy company - every time I try to work without Opera I get crazy.
I also hope they continue to be healthy, but currently asking web developers to support 4 totally different browsers is getting somewhat unrealistic, especially when one of those only can claim a sub 0.5% at best average market share, there's no way you can justify the debugging expense and time, I could but won't post a code sample of very simple css 2 that Opera totally fails to render correctly, a huge margin bug, I had somewhat naively assumed that it would finally run in 8, but of course it doesn't. Very tiresome, that's a showstopper bug too, destroys layout.
I pick the top ones now, by market share. Anything under 1% is only supported to get rid of show stopping rendering bugs, no fine tuning any longer, in other words, MSIE engines, Gecko engines, and Safari/KHTML engines. After that it's just whatever I feel like putting time into, usually links/elinks because I like those.
I avoid most css that I know to be a show stopper in any major browser, which means I work with a restricted css pallet. Obviously all css and html validates, this is not, and has never been, a validation issue.
|I could but won't post a code sample of very simple css 2 that Opera totally fails to render correctly, a huge margin bug, I had somewhat naively assumed that it would finally run in 8, but of course it doesn't. Very tiresome, that's a showstopper bug too, destroys layout. |
There are equally annoying bugs with Firefox, not to mention the minefield of layout bugs in IE6. Have you brought the Opera bug up on their forums? Have you reported the bug directly to Opera?
Regarding stats, something seems wrong here to me. There are millions of Opera users worldwide, yet how come the browser always features so lowly? Are a lot of stat programs simply not testing for "opera" in the string and assuming they are seeing IE or Mozilla? Or is that in terms of the web, even millions of users is a drop in the ocean?
As I indicated, testing sequentially for a substring within a string is not difficult, and no stat program I've tested has any problems with correctly identifying opera numbers.
Why millions have been downloaded but stats show average counts of around .2% isn't an important question to me, I know the stats I'm looking at are correct, I, and others I know, have tested them, they are 100% accurate.
I'd say the market has spoken, and continues to speak, I've always followed what the web really is doing, even when it's not what I wanted it to do. And what the market said, and continues to say, is that an ad supported browser that costs $39 to remove the ads is not of interest to any but a very few power users who like those specific features that Opera has, people like you and tedster.
Posting a bug report on opera would take more time, and as I indicated, I'm not interested in spending more development time on Opera, their market share is simply too low to justify any development time at all. When I finish a new layout, I check it quickly in 8, sometimes 7, just to make sure it's more or less ok, and that's it.
When Opera was the number 3 choice, and it seemed likely that this market share would improve, I spent a lot more time supporting it, now it isn't, and the market share has remained fairly steady, and shows no signs of changing, so I don't.
Given that Haakon is supposed to be the big css guy, I'd hope that he and his team are already aware of the bugs present, and if they aren't, it's not my job to report them, again, I'm more interested in helping out various open source projects if I'm going to spend time working out bugs for someone, I'll spend some of my time to help people who are spending their time helping me, it's very simple, basic open source ideas.
I take this very pragmatically: I have to create good sites that work in most browsers, MSIE demands learning how to deal with their many css bugs, due to their market share. I'd rather just ignore it, but then clients get annoyed at the errors. I know of one minor Firefox bug that's easy to work around that I actually run into on real layouts, I'm sure there's many more obcure ones, I just never find them.
If I ask a client if they'd like to pay to get better support for .2% of the market, or less, their answer is no, they don't care.
The bug I'm referring to, and I don't want to really go into this question since it doesn't have much to do with the topic of this thread, blocks the implementation of a very standard layout, or a layout that would be standard if all the browsers correctly supported it. Currently only Firefox, and maybe the new code fork of apple webkit's khtml implementation, supports that method, which is very simple. It's not some obscure css thing that only css freaks would ever be interested in, it's a major failure, I was quite surprised to see such a failure in Opera, but given that 7 didn't even remotely support that feature, at least it's partially supported now in 8.
To me it's clear that Opera missed finding the sweet price point that would have moved far more units, and generated more revenues in the process. Their inability to understand this simple business reality year after year is another thing that makes me completely uninterested in their product, they simply don't seem interested in expanding their market share, so I'm likewise not interested in supporting their miniscule percentages. Which is not to say most of the stuff I code doesn't work in Opera, it does, usually anything that works in MSIE works in Opera.
Another example of the human mind being unable to understand coincidence...
So what there were no Opera hits during the month, this does not mean people aren't using Opera, or using it more or less.
If you value one group of people using one browser over another all you can do is target the search engines that are aquanted with the default pages of those browsers though it seems silly to me.
A comment may strike the Earth once every tens of thousands of years however mankind has barely existed for thousands. When something normal happens that in the human sense of short-term that usually happens on a greater scale everyone freaks like it's some religious armageddon. These things happen all the time and it's part of the nature of existence.
My two cents is that you should relax, as long as it works in the big four then you've got your basses covered.
"So what there were no Opera hits during the month, this does not mean people aren't using Opera, or using it more or less."
Actually, it does in fact mean exactly that, obviously. And what's more, the number of opera hits I got 4 years ago are statistically almost exactly the same, this month, last month, and the month that will come, they never change, except when they go from 0.2 or 0.1 to 0.0. Those hits do not change from site to site, topic to topic, except as noted when the site caters to a technical audience. Every new site I take has the same rough numbers, from 0 to 0.2% at most. Doesn't matter what subject area, it's always the same.
This is one of the more odd pieces of logic I've come across here, web stats tend to be very reliable because they are constantly in flux, people come and go all the time, different users, different tools, makes for an almost ideal way learn about what people are actually doing out there.
That's why the original poster made his comments, he'd noticed a small movement downwards in the average numbers, presumably across different sites he runs. I also run different sites, lots of them, different industries, but the averages don't change at all for non-technical sites, never go above 0.25%, usually are lower, closer to 0.0%.
When you then take each webmaster's numbers, figuring in what type of audience is attracted, you get very accurate results, but for some reason people just won't believe them, well, some do, some don't.
He is talking specific to his domain(s) based upon general statistics be it either his historical data or the data of other sites.
Relativism is not proper to use in this situation as we're not talking about basic concepts used to identify objects when creating a language or talking about highly abstract concepts. Each site has it's workings and place in existence and the fact that he didn't get any Opera users does not mean everyone all of a sudden stopped using opera.
If a comet hits us then many people will think it's the end of the world and religious ideas of armageddon are coming true even if say it is normal for one to hit every 50,000 years.
Perhaps the concept of determinism is too complex for some to understand without a full fledged discussion. Anyway the Opera stats are likely to return to normal or near normal.
Just because something is not probable does not make it impossible.
No.. it's just holding its own! (user-agent-strings taken into account)
all this discussion does take me back to the good ol' says whan the only argument was should we even try to support NN4 ;)
HatTip: It was Opera who pushed limits (even though it was only geeks and power users that used it) but thanks to them we have a wider browser market and thankfully better CSS support, today..
All FF did was to do what Opera did but marketed it better amongst the General Public.. (I know I have had non-pc folks ask me about it, which is enough.. considering they never even heard of IE most of the time ~ "The blue "e" yes but what's IE? ;))
As ever if you're building sites that you want Opera to support then go with the latest version for testing.. if (sorry Ted) a die-hard user chooses to stick to an older version because they like it, rather than upgrade like the power users they are, then I fear they will be the ones to miss out :
Very much like why we have to put up with (and shut up about) IE really ;)
JAB Creations, you're thinking too hard about this, web stats due to their randomness form predictable behaviors, when encyclo says he has noticed a drop on his busiest site that's what he is seeing, if the visitor numbers are high enough, which is relevant, since if you're getting less than say 1000 uniques a week the small number of opera visitors you'd get would literally be the difference between 4 or 8 people visiting your site with Opera, but once the numbers go higher than that, you're not talking about individual behavior any longer, but group behavior. From reading other posts encyclo has written, I tend to pay some attention to his observations. Encyclo, what kind of visitor numbers are you talking about here?
Re Opera in the wider sense, if the web had worked the way I wish it had, Firefox would be number one, Opera number two, Safari/Konqueror number 3, and the rest would occupy their niches. If Opera numbers on the desktop would rise a bit, those annoying bugs they keep having would magically vanish, since there'd be more eyeballs on it, and more motivation to take care of them, currently that motivation is to support handhelds, as tedster noted, and it shows.
Also worth keeping in mind is that until last year, it was reasonable to expect more NS 4x users than opera users... still happens, not often though
OK SuzyUK, I've been shamed into upgrading to 8.01. It was painless, as usual with an Opera upgrade.
|So what there were no Opera hits during the month, this does not mean people aren't using Opera, or using it more or less. |
I am assuming nothing, that's why I asked. ;) Stats programs are good at showing trends if not specifics - as I consider that the site in question is already non-typical (with over 7% Safari and Firefox only at 5%) I simply wanted to see whether others had noticed a downward trend in Opera hits.
I looked at a more technically-oriented site today and the trend there is definitely downward too, but only slightly. I doubt that Firefox has had much of an impact of the paid-up Opera community, but I strongly suspect that Firefox is taking the more casual user who is not prepared to pay $39 for a browser. Firefox has taken a huge share from Mozilla and is making inroads with the non-technical crowd - something that no other browser has managed since Netscape was strong.
|By the way, the one browser that can avoid stat programs completely is Safari/Konqueror, where it has the option to give a user agent string that is an exact duplicate of the MSIE XP string, that's impossible to detect |
You can detect the browser type (not version specific) via other methods than the user agent string, but as the user agent spoofing is not usually statistically significant (as GaryK says) it is not worthwhile doing other than as an excercise in scripting.