| This 61 message thread spans 3 pages: 61 (  2 3 ) > > || |
Recent changes in the alexa traffic ratings system
In late September Alexa introduced a new set of traffic reporting displays for all the sites it lists. Traffic can be shown for 3 months, 6 months, and 1 year. The traffic data is collected from those using the Alexa Toolbar for Internet Explorer.
A composite traffic ranking is shown on the toolbar when you visit a site. Clicking on the Alexa rankings number, will take you to information about the site you are visiting.
I wanted to bring up the topic because I feel the traffic measurement by the Alexa toolbar is one of the most powerful and accurate measures of internet statistics I've seen yet.
Many of the other net ratings systems rely on sampling a group of selected users. Often, those users put something in between their machine and the internet that tracks what sites they visit. That is highly accurate for those people, but the sample rates are so low that the data is suspect when applied to the world wide audidence of the net. Some of the net ratings firms use sample groups of only one thousand to two thousand surfers. I feel they are statistically invalid. If it were anyone but the companies behind those ratings numbers, they would be discarded as noncredible.
The Alexa toolbar on the other hand, has a reported install base of several million. That is a significant group of people. Even so, Alexa does warn that certain sites and sectors of the net could be under reported or over reported. Factors such as Alexa's own site where a high percentage of users will have the tool bar installed, or linux heavy sites where few will have the toolbar installed will be over or under reported.
In the past, the Alexa Toolbar has been a source of controversy. Much of that upset has died away with the acceptability of tools like Netscapse "What's Related", the Google and Yahoo toolbars. When you visit a site, the url is transmitted to the homebase in order to return the site information.
Alexa is also connected with Archive.org [archive.org] (Wayback). You can click on a button to see the history of any page.
There is also an "auto site suggestion" while you surf. The Alexa Toolbar Related Links are provided on the toolbar to related sites. I think this is one area where they need to do more work. It has to be a fairly high traffic site in order to become accurate on the suggestions.
The search feature of the alexa toolbar is powered by Google and can be installed at the same time as the Google bar. I think the traffic figure is better at determining site quality than Googles page rank. But taken together, it is a powerful synthesis of data available while surfing.
Alexa Info for webmasters [pages.alexa.com]
Download and install the Alexa toolbar: [download.alexa.com...]
Of course, if you cannot - or are not inclined to - install the toolbar but still want to see their stats, you can always just visit the Alexa home page and search for a site name.
I'm not inclined to because I already have the Google toolbar installed so I can see PageRanks. It's possible they can co-exist peacefully but I'd like to keep _some_ of the window free for the web page.
And I think I'm right in saying the tool bar needs Windoze and IE5 or above.
I would absolutely agree that the mainstream traffic estimation services are based on very small samples, but still to be convinced that Alexa data, though based on higher sample sizes are more representative of the web site browsing population.
I also note that the mainstream services collect data on US uses mainly I think. (Correct me if im wrong)
However Alexa ranks some East Asian sites very highly suggesting an over-estimation of say Korean sites, - and the other major biases are well covered by Brett's first post here. Finally it takes a special type of person to install a toolbar, you would need to be on a high speed connection to put up with the extra downloads involved each time you look at a page, you need to have some internet savvy, and particularly interested in getting detail "credibility" data for web sites you visit. I believe that this represents a bias away from the casual mainstream user.
A quick review of our own sites shows the "ranking" to be a bit off base, with some of our less popular sites, being ranked as higher in traffic than our most popular ones. In some cases the information Alexa provides is a year or so out of date.
The related sites were ones we suggested ourlseves around 3 or 4 years ago when you could "vote" for related sites.
Our sites are not in the top 10,000 by a long way, so maybe Alexa only "works" for the real "players". However i still find it hard to belive, despite South Koreas massive web surfing population that South Korean portals could be so far up the top with a sample even approaching enough representativeness to justify generalising from.
Portents for the future are not good either! People are already discussing how to get ranks for their own sites up by installing the Alexa toolbar.
We already know that a sign that a Web service has reached popular credibility and popularity status is when it is seen to be worthwhile spamming ;)
Our site uses the Alexa rankings as sales tools to compare all sites in our category (Japan). We have been using this system since September.
The daily chart is especially useful when adding new features to your site. The graph is two days delayed so you have to wait for results to show.
You can also check the movement of all your competitors and even overlay the graph with that of a competitor.
For the first time the numbers behind the ranking system are shown, which gives the system more credibility. I suspect that, based on these numbers, an estimated unique users chart will be coming soon.
You can also see a percentage listing of subdomains giving useful information about which parts of your competitors sites are popular.,
We added put all of our domains under one root domain to increase our ranking which went from 20,000 to about 12,000 now and higher from next week. We are aiming to beat our rival, which has 11,000 or so.
We can show this information to advertisers and to the press as independent proof that we are growing and that we are becoming the leader in our field.
The thing that I find strange is that Alexa uses Google to tabulate its search results. Surely they should use their own search rankings as it is more accurate than pagerank. Alexa rankings are based on user popularity not webmaster popularity.
In our category we have the same page views as our main rival but because they have more incoming links they are second place in Google. We used to be tenth, but are now down at 30th for a reason which I cannot figure out (we have daily changing content, more uses of the Japan keyword and more incoming links than many of the sites above us.)
All in all I think Alexa is an excellent tool for monitoring and comparing your site against your competitiors.
I used the Alexa Toolbar as a lark, but quickly got hooked on it. I think the traffic ratings are far more telling about a site than Googles PageRank figure. The google number is about the page, while the Alexa number tells me about the entire site.
Alexa Toolbar hasn't replaced the Google toolbar for me, but it does compliment it nicely.
Like Googles toolbar, it transfers the url back to Alexa to get the traffic figures. All toolbars of these types do that. Not my definition of spyware.
Just checked 6 months of logs for 8 very different topic sites and in every case IE browser was over 95% (2 are at 98%). Pretty good sampling don't you think? Why, if you are interested in site stats, are you so concerned with the 5% which is unreported?
This is not meant to be rhetorical; I just regularly hear people at WW put much weight on tools and browsers that are lightweights in regards to significant and useful numbers for website strategies. What am I missing in these smaller numbers?
I pretty much throw out the smaller-numbers-browsers and search engines. Don't you only concern yourself with Yahoo, Google and MSN for search engine results, for they make up 99% of searchers... do you really spend much time on the other se's? Are we not trying to build for the mass?
I don't know how accurate Alexa's traffic rankings are these days, but I do find one thing a bit odd: When I look at the "1 year" graph for my site, it shows a ridiculously high ranking for a year ago (when my site had just been launched and had very little traffic). The graph starts off around 20,000 in October of 2001, drops off the charts, then comes back up to my current ranking (which is typically somewhere in the 40,000s).
The daily ranking swings also quite extreme. A few weeks ago I fell to 200,000 one day; a day or two ago, I was at 20,000. My server logs don't show any extreme variations, so either Alexa's plotter needs better damping or every other site on the Net is jumping up and down enough to create big swings in rank for my site.
Still, it's fun to play with Alexa. I can ignore the bad days and enjoy the good days when I outrank Travel & Leisure Magazine and other big names in my field. :-)
yes and no, of course I'm trying to build for the mass, and i hate building for NS4.x (so I don't do it anymore) but I'm also not building for the us-market but for the german, and here IE has 'only' 84% and I also try to go for the other 16%...
Also that thing with spyware is a really BIG issue. people here are very concerned about their privacy, i think much more than the americans are. are there any numbers about alexa stats and europe?
>In Europe, is IE gaining or losing market share?
Not possible to deduce. All the newer browsers fake ie agent names. Within a year, I think ie will have 100% of the browser market just because people have to emulate the agent name for compatability with ms's proprietary server (.net) products.
The antitrust case is over. The monopoly won. Stand back - they are coming through.
|The antitrust case is over. The monopoly won. |
I'm not so sure this is a bad thing. For nearly 100 years AT&T had a monopoly on the telephone service in the US. Because of that monopoly, any phone can call any other, and phone service is compatible around the world. I don't have to worry if my neighbor is using Sprint today or MCI - I can still call them.
This business of worying about coding for Mac or Windows or Linux, or swapping files betwees OSs, and wondering about which browser people are using is frankly a ridiculous waste of creative time and energy.
Competition is of course good. Without Microsoft competing against Netscape for browser wars, we'd all be still stuck with NN 4.X as the standard, and I think most people agree that would be atrocious.
But we are still in the infancy of the internet and personal computers in general, and just like in the infancy of other big industries there will a shake out and one defacto standard will emerge.
>In Europe, is IE gaining or losing market share?
Internet Explorer 70%
AOL Browser 8.5%
Fresh from a definitely non techie site in german language.
I guess it's nevertheless fair to say Alexa being restricted to IE dosn't distort the stats too much. I'd suggest the problem with Europe being rather nobody knows the Alexa toolbar.
|I'd suggest the problem with Europe being rather nobody knows the Alexa toolbar. |
I agree with you. On the contrary that Google toolbar, which I know many non-techie users who have installed it, Alexa is almost unknown by the European surfers (at least Spanish surfers). Also, I think Europeans are more concerned about spyware (kazalite success prove it)
Anyway, I've been looking at the statistics from my site, and the say our Traffic Rank is going down, although our traffic has increased about 50% in a month (from 35K daily page views to 55K)
Another last thing. In our statistics "People who are interested in this site also bought" all the books are related to carpentry or "do-it-yourself" (although our site is not related in any way to that hobby). I deduct from this that there're not a lot of surfers who visit our web and have the toolbar installed (basically because it's in Spanish).
> Traffic Rank is going down, although our traffic has increased about 50% in a month
Traffic rank is relative global usage not specific site performance.
Like criscross mentioned, Alexa is the greatest tool when it comes to comparing traffic with your competition, since the surfer demographics is the same!
We are having an extremely tough time convincing our customers that the "big name" competitors actually get less traffic than we do, though they charge 10 times as much for advertising.
Our competitors also seem to be claiming significantly higher pageviews than us. Either they are lying or Alexa is! I yearn for the day when web traffic audits become reliable and standard for the sake fo advertisers.
|I'd like to suggest another significant flaw in the Alexa scheme. Its unlikely that any school lab, library, etc. (any public access type location) would have or allow users to add a toolbar of any type. Thus any users at such a venue won't be counted. |
That could be true of many corporate settings as well. Still, Alexa compares favorably with rating services like Media Metrix (which uses a much smaller sample, monitors fewer Web sites, tracks only North American users, and provides statistics only to subscribers who can afford its high fees).
I think Alexa Traffic Rankings can be a useful tool if you're comparing apples and oranges (i.e., sites in the same category) and you're looking for ballpark information rather than accuracy that's on a par with server logs.
I use Alexa and find it very informative for finding similar sites as they are grouped by category. However as for the numbers, I am not overly sure how accurate they are.
Basically it all boils down to how many Alexa (people with the toolbar) users visit your site. If you have a lot of Alexa users who spend or re-visit your site frequently then your rank will be very high.
I just use it as a guideline along with googles pr.
ADD IN : Mind you it is addictive in a sense as when you are at 30K you want to be at 20K then 10, then 1 etc etc!
In Scandinavia we have alot higher percentage of IE users. Some of my clients, that target a broad spectrum of users have less than 1% NN users. So for us the Alexa numbers are fine.
Also, it looks like at least the Danes do use the Alexa toolbar as one of my clients (purely Danish content) have a Trafic Rank of 258. Not bad for a local Danish site (and they still want more!)! :)
>comparing traffic with your competition,
>since the surfer demographics is the same!
Totally right on!
I agree, there is a cutoff point where the data becomes suspect because of the lower traffic numbers. Seems like about the top 5k sites are pretty accurate and the top 2k sites are extremely accurate.
It's just that this is the only tool available that ranks sites based on traffic. It leads to some dramatic results.
Go check the top 100 sites. It's startling to see what sites are REALLY getting traffic. There are more than a dozen sizes in the top 100, that I have never heard of. That's because those sites don't show up on the lower sample rates of the other ratings systems.
I believe this tool is rewriting internet history as we speak. It blows the doors off the myth known as Nielsen net ratings.
I found this post interesting. I have 6 websites and in all of them the traffic is completely different from Alexa rankings.
It is very simple to verify how alexa can be easily manipulated.
Just look at your traffic before and after you install alexa bar.
I was able to improve 500000 positions just with my normal daily visits to one of my websites. And I have the toolbar for just 2 weeks! This website is around the 100000 in ranking and it is projecting from the daily ranking to jump to 15000.
Bottom line, Alexa ranking says absolutely nothing. Off course, with the exception of the big players (first 500 or 1000 websites).
However, even in the top 1000 you are able to find some great disparities. Specially among Brazilian websites, where I am from.
Yes, I wonder where that cut off point is.
Do you think that for many of the small biz sites that get maybe 5k - 20k visitors annually the Alexia Traffic Rank will mean much?
I think the next version of IE from Microsoft will have something on the lines of Alexa or, it wouldn't surprise me, they'll tie up with Alexa and bundle it with IE. That's when it may not hog up as much screen real estate as it does now. It will then have real influence and its rankings will really reflect usage, over a period of time.
Right now, it is more a tool for power users, not for the casual surfer who may not be able to even install it.
|looking at the results from our site we currently have a reach of 100 users/day. I wonder if this means that 100 Alexa users have accessed our site in one day.|
According to statistics theory it is possible to accurately predict large populations to 95% accuracy (+ or - 5%) with only a few hundred results (about 384). ie with a sample size of 384 we can make a survey that if taken multiple times will have a 95% chance of acheiving the same result each time within + or - 5% difference in the ranking each time.
Over time the results should be very accurate. I think that is why the three-month average is used as the "public" ranking.
384 divided by 3 months (90 days) should mean that a reach of 4/day is accurate for a three-month time period. That is anything less than a ranking of about 150,000.
However, I am not a statistician. Are there any stats people out there that can give a better analysis?
I totally agree with you. If microsoft bundles alexa in ie, then we will see, for the first time in internet history, a real and accurate traffic mesure.
I am anxious to see that happening.
I thought Alexa was built in to a previous version of IE (maybe 5.0) and removed later.
|looking at the results from our site we currently have a reach of 100 users/day. I wonder if this means that 100 Alexa users have accessed our site in one day. |
Nope, that means 100 in one million users. So, one person in every 10,000 web surfers visits your site.
|According to statistics theory it is possible to accurately predict large populations to 95% accuracy (+ or - 5%) with only a few hundred results (about 384). ie with a sample size of 384 we can make a survey that if taken multiple times will have a 95% chance of acheiving the same result each time within + or - 5% difference in the ranking each time. |
Statistical extrapolation from a very small sample size doesn't apply to the web because there are so many web sites out there, and there is also a very diverse array of users. It is not just 60+ cable TV channels we are talking about here, there are millions of web sites on the Internet.
|I think maybe Brett just wants us all to get Alexa installed so his rankings go even higher! When I joined in May 2001 W/world was at 2800 or something now 889 and still going up! |
I don't know if this is the case, but honestly Brett, you do seem to be hyping this as the greatest thing since sliced bread. I personally don't see it as being all that accurate anyway. Nor do I find it terribly useful to see the top 100 sites as there is little chance I will ever have good links from them. Cute, interesting, fun to look at - sure. Useful? Hardly.
Alexa rankings have always struck me as a very rough approximation. They are far from scientific. Anyone with a basic background in stats could explain why there are huge holes in most current methods of measuring web site traffic.
Call someone up from a company like the Audit Bureau of Circulation (or their online division which just got downsized) and ask them for a list of the methodological shortcomings of the major panel-based measurement services such as Media Metrix. I'm sure that they'd tell you that Alexa was *even more unscientific* than the seriously flawed methods used by major web metrics services.
A large installed base means little, IMHO, because it is not a random sample...indeed it may be an extremely unrepresentative sample... and in addition, owners of some web sites may indeed deliberately go out of their way to create false visits.
If it only takes a few hundred Alexa visits to make a big difference in the rankings, and a web site decides to get a bunch of their friends together to skew the ratings... well, you can figure it out.
I agree, it's a rough way of comparing to see if sites differ by orders of magnitude .... but I wouldn't go gaga over it.
|I personally don't see it as being all that accurate anyway. |
i find it rather accurate. the absolute numbers aren't too correct but the developments are. it managed to jugged the drop in traffic of my site after its move "from free to fee" very accurately. and it is the only place (i know) where you can more or less evaluate the size & importance eg. of a future cooperation partner.
muesli I am not sure that is entirely true as it does not log visits but only visits by people with the toolbar. Now how many average users have the Alexa toolbar installed?
Looking at Googles attempts to get more an more people on their system I would say that it is a tough sell. The only people I know of that have it installed are webmasters or such like!
| This 61 message thread spans 3 pages: 61 (  2 3 ) > > |