The Great Alexa Spike of 2006

Tweet

It seems to have passed without much comment, but there seem to have been some significant changes at Alexa recently — most notably sites that appeal to a more tach-savvy audience are making huge leaps up the rankings. The phenomenon can be seen clearlyacross all major web development sites — from W3Schools.com to W3C.org to SitePoint.com — but also across a long list of sites with more general appeal to tech users, such as Flickr.com, Del.icio.us and Slashdot.org.

Meanwhile sites with no notable ‘tech-skew’ (i.e. CNN.com, EBAY.com, etc) have either held firm or been shuffled backwards by the sites bubbling up around them.

It raises some interesting questions:

  1. Have Alexa changed their statistical algorithms?
  2. Is this a temporary anomoly?
  3. Which sites lost out the worst?

Alexa Rankings

It certainly seems that the Alexa ranking system has undergone a radical shift in it’s demographic in the past 18 months.

Alexa got their first big push in 1998 when they had their toolbar included in the Netscape 4 default install. A year later they pulled off a huge coup by getting it included in Internet Explorer 4.

At the time most users would have just assumed (as I did) that it was simply a part of browser. With a reported install base of around 3 million users (a lot in 1999), this was probably the broadest sample audience Alexa ever had access to.

In 2000 Amazon purchased the company, and at that time Microsoft chose not to included it in their newest browser release – 5.0. Users could still have the bar, but they had to find the site and install it themselves.

This is the point where I think their userbase started to change. More technically-advanced users moved quickly onto IE5.0, and at the same time were becoming more sensitive to terms like ‘spyware’. Alexa’s modus operandi of diligently reporting their surfing habits back to their servers didn’t sit well with their ‘big brother’ fears.

As a consequence, the Alexa userbase seemed to take a skew towards a less tech-savvy user, users who felt they got real value of the tools Alexa provided (‘What’s Related’ links and the Search tools in particular). In essence, while AOL users might well use the toolbar, Slashdot users wouldn’t have dreamed of it, and that was reflected in their relative rankings.

That’s where the situation stood well into 2004, until Firefox started getting some serious traction — particularly at the ‘pointier end’ of the tech audience. Although those guys never ‘dug the spying thing’, many web geeks still attached some value to the pure ranking figure. Sure, it wasn’t all that accurate, but it was better than anything else available and so became one of the key metrics used when buying and selling web sites.

It’s a case of ‘if enough people believe something has value, it actually does‘.

And what was the easiest way to keep an eye on those Alexa rankings? Why, extensions such as ‘SearchStatus‘ that provides real-time Google PR, Alexa Ranking and many other useful metrics to site users, owners and developers in a handy status bar format.

However, the easy bit to forget is that when the toolbar tells you a site’s Alexa rank, it had to first tell Alexa’s servers where you were at that moment. In other words, you became part of Alexa’s statistical pool.

This has signalled a major shift in Alexa’s demographic, and would seem to explain much of the reported growth across this section of sites. As Alexa’s newbie userbase has gradually decayed, tech users now represent a very statistically significant percentage of Alexa’s grunt.

It’s also worth noting that although Apple’s audience might be considered relatively tech-savvy, their preference for Safari rather than Firefox (and it’s Alexa powering extensions) seems to have insulated Apple-related sites from this constant rise — including the recent spike.

This is, of course, all just my personal speculation and I have no extra access to Alexa’s internal methods and data.

It also doesn’t explain the spike.

Could it be…

  • A change in ranking methodology?
  • An enforced upgrade of either Firefox or an extension that made new data available?
  • Alexa deliberately excluding some sites for the ranking?

Those are my best attempts, but maybe you have some other ideas?

Free book: Jump Start HTML5 Basics

Grab a free copy of one our latest ebooks! Packed with hints and tips on HTML5's most powerful new features.

  • marcin

    Have you taken Easter Break into account? :-)

  • John McL

    Have you taken Easter Break into account? :-)

    ranking system has undergone a radical shift in it’s demographic in the past 18 months

    In the UK we only get about 2 weeks max for easter break =s

  • DASH
  • adholes

    I own two sites that get a relatively good amount of traffic: Adholes.com, which is a network for ad industry people, and Punknet.com, which is for punks. When I started Adholes and only have about 400 users, I had traffic spikes on Alexa that were much higher than the ones I get with 5,000 users and 4 X the page views I had a year ago from my own stats. Furthermore, Punknet also has over 5,000 users, and in my stats, regularly schools Adholes in pageviews because of its teen audience. So how does Alexa rank us? Adholes hovers in the mid 200,000’s. Punknet is down in the 4,000,000 somewhere. WTF? I know for a fact that Punknet gets more traffic. So, on that note, should I even be happy in the occasional times Adholes has cracked 50,000? Or is it just the inverse of the unexplained anomoly that has had Punknet buried. My best guess is that punk kids are savvy enough not to have the toolbar, therefore no Alexa love. And that as my site has grown over time for Adholes, more and more users have abandoned it as well?

  • cstegner

    adholes, thanks for the post. It now forces me to not have full belief in the statistics I once held so dearly to my heart.

    So if Alexa.com is that far off base, what is our best tool for understanding our position to the rest of the internet?

  • http://www.sitepoint.com AlexW

    What did Churchill say?

    It has been said that democracy is the worst form of government except all the others that have been tried.

    I think something similar holds pretty true for Alexa.

    Adholes, while I agree your punk audience would be unlikely to install the toolbar, I also think it’s highly likely many of your Advertising network users would be running the ‘SearchStatus’ extension, and that would explain the high rank and the spikes.

    In real terms, I would guess there may only be 3 or 4 users in your punk audience with either the bar or the extension. You may only have 500 advertising users, but I wouldn’t be surprised if 50-100 of them had the extension installed — if you’re buying or selling advertising inventory Alexa data is a tool of the trade.

    See, Alexa always tends to look it’s worst when you’re comparing apples and oranges (advertising guys and punks in this case) but you generally don’t need that sort of data. Most of time you want to know whether to purchase ‘Site X’ or advertise on ‘Site Y’, and what you really need to know is how those sites compare with their direct competitors.

    My view is, if you compared adholes.com against a competing site with a similar demographic, you’d most likely find the data more credible and more useful.

  • adholes

    Alex, great point. Just to clarify, both sites have around 5,000 users, but Punknet has much more traffic. And I only wish we had ad sales that people wanted to track. If anyone can figure out how to sell ads to ad people, please let me know! :)

  • Anonymous

    Do you think it could be related to Google Ads?

  • http://www.sitepoint.com AlexW

    Doesn’t seem likely. Couldn’t imagine that Adsense would be more common on tech sites than non-tech sites

  • soomp

    wow – i just discovered this site today, and i was wondering the exact same thing! check out my site stats. there’s this huge and unexplicable jump right around easter (mid-march). at first i thought i must be totally awesome or something ;P, but something must have changed in their system. i’d be interested if anyone figures this out! :D boss at soompi dot com.

  • http://www.sitepoint.com AlexW

    soomp, you’ve actually provided an interesting spin to it. Comparing graphs, your spike starts about 4-6 weeks earlier than the others, so it’s probably initiated by a different source.

    I’m wondering if maybe something like a Korean language version of the toolbar or extension was released recently?

  • 3PointRoss

    I have never been completely convinced the benefits of Alexa and Alexa ratings. It seems it is too easily manipulated and its ratings doesn’t directly reflect the real world situation of a website.

  • Pingback: BlogCadre

  • franco78

    I thought I was the only one wondering what Alexa is been up to.

  • Brad

    Has anyone looked into Hitwise, they charge a bucket load for their service but its nearly 100% accurate with rankings. They actually buy the log files from ISP’s all over the world, doesnt get much more accurate than that.