browser stats/why we love the numbers
The beginning of a new month. The (modest) pile of cash on the bank is a little more substantial and the fresh slate idea softly settles down inside our minds. As web developers, this is also the time for browser stats updates. The importance of browser stats have been under heavy attack, but for those who know how to interpret them (and those who just love numbers), they remain are a valuable tool. This article will take a closer look at what they don't, but more importantly, do represent.
why they matter
Of course browser stats are pretty important to browser vendors and fanatics alike. Webkit fans like to see how much the engine's market penetration has risen, FireFox people are curious about the adoption rate of the newest FireFox version and the people of Microsoft are probably scratching their heads when they see how well ie6 is keeping its ground.
For web developers, it's important to see what browsers (and versions) we need to support. When a browser drops below a certain percentage we can choose to stop supporting the browser, saving us some testing time. Rest assured that all css people are checking browser stats to see a sudden 20% drop in ie6 stats, sadly this is only wishful thinking.
Bottom line, browser stats tell us something about our job which can be useful when developing websites. So why the negative attitude?
pitfalls
The problem with browser stats is that they are not very reliable. There are various factors that influence the stats to render them quite useless. A quick rundown of the three most influential issues.
- First problem is that most stats are based on visitor statistics on a very limited range of sites. This range of sites is not determined to get solid statistics, but is usually determined by site ownership of the stats owner. Simply put, you might get very different results from a party that only measures European sites in comparison to American sites.
- Another problem was introduced by several browsers themselves. Browser stats are based on the user agent string, which can be altered by users themselves (a fix that dates from times where some sites would block certain browsers). So you could very well be visiting a site using an Opera browser, but you'd still end up in the stats as ie6 user.
- Finally, not all visitors are human visitors. There are many spiders and other automated web visitors crawling around the web and most of them identify themselves as ie user agents. Sadly, they are of very little importance to the stats.
As you can see, all these factors influence the stats and distort the actual percentages.
conclusion
Even though the percentages given on browser stat sites are not very definite or trustworthy, they are still interesting when compared to stats from earlier months. The actual percentages might not be too definite, but the trends can definitely be of interest to us. Besides that, as you lay stats from different sources next to each other interesting parallels will pop up. For example, it might be hard to pinpoint the exact percentage of ie6 and ie7 usage. On the other hand, since a short while all browser stat sources indicate that ie7 uses is larger than ie6 usage, which might accounts for something.
So don't just throw away browser stats, but look at trends and compare other sources the find some decent information about global browser usage. It might help you when defining the scope of your work.
sources
For those interested in online sources for browser stats, these are the sites I'm using myself.
- marketshare stats (benchmark statistics)
- w3counter stats (with an indication of stat sources)
- w3schools stats (web development crowd)
Its also worth pointing out a few other issues with these stats. As you suggested, Opera may not identify as Opera due to days when sites blocked certain browsers. This is still not past tense as sites still block us, and we have to add spoofs for certain major sites to avoid getting blocked. Site sites are often also very US or anglophone centric, so that gives a fairly biased view against browsers like Opera that don't have a strong user base in English language countries (Opera's strongest markets are countries like Russia, and Indonesia for example, and large parts of central and eastern Europe and Central Asia, and growing quite strongly in South East Asian countries).
The issues not mentioned about stats is they are often on page views. This is often why mobile browsers don't figure very strongly, even though we know that Opera Mini has a much bigger user base ratio to Opera desktop than stats show, where they suggest Opera has around 1% or lower while Opera Mini is 0.05% or so. Users on mobile usually visit less pages per site for bandwidth and other reasons. Browsers that cache agressivly will also show less hits per user as they will be hitting cache more often instead of creating new http requests. Safari on iPhone is particularly guilty of this as it has a tiny cache. Opera on desktop tends to use more cache than most browsers. There are also browsers that pre-fetch pages that were not requested, to speed up page loading when/if a user does access that page (and example would be pre-fetching the next page in a Google search result and the top one or two results).
As an example of what you don't see in stats (unless you are based in Russia, at a high traffic site), is that Opera has around 20% or higher marker share in Russia, Ukraine, and other former Soviet Union countries.
Even stats for your own site can be misleading. When contacting sites that block Opera, I'm often told that according to their stats, Opera has less than 0.01% marketshare on their site (from their own stats). What they fail to realise is that if they are blocking a browser, it is automatically going to have a low market share as users can't actually access the site to show up in the stats. They'll hit the block page once and either mask that domain if they are tech savy or not come back if a regular user.
Hey David,
thank you for clarifying some of the more obscure issues of webstats. It's sad to see there are still sites out there actively blocking users using a certain browser. Since you seem to contact those site owners, can you say anything about their reasons for doing so? Can't come up with anything myself.
As for the stats, Opera is quite known for getting a hard rap when it comes to browser stats (often below 1%) but has deserved its status as Grade A browser over the years. I always test in Opera, though I admit that in a Safari - Opera bug battle (somehow they arise on every project) Safari gets the fix.
Keep up the good work though, I really like the improvements made in 9.5 and as long as Opera looks alive I'll keep checking if sites run okay :)
It varies. Sometimes it is just “you don't have enough market share”, sometimes it is because they don’t want to spend QA resources and are not willing to just allow unsupported browsers to give their best shot at working anyway. Sometimes it is malicious, and sometimes it is because Opera didn't work on that site, and instead of looking into why they decide to block. If the site works in other standards based browsers, say Firefox and Safari, the latter case isn't too often.
Well, that makes for an interesting read. I guess convincing them of their silly reasoning doesn't quite cut it, or else you wouldn't be in this mess :)
That "malicious" remark is one to remember, luckily people like these are disappearing, though slowly.
Hopefully Opera might become a bit sexier on the Mac and pick up browser share with Hicks being part of the team...
I have recently seen an example of a web developer quoting stats as a reason for not making a site work with IE6 (and because the browser is more than two years old and IE8 will be out soon). All seems a bit flimsy to me, and made worse by the fact that they propose to use a "your browser is out of date" message.
* required fields