Alexa Says YouTube Is Now Bigger Than Google. Alexa Is Useless

Alexa was doomed when it required the toolbar for its statistics. It’s way too easy to manipulate, and honestly who do you know that uses it besides people looking to increase their alexa rank?

Alexa has not worked right in many years. Its no where even right.
I just recently discoved your blog and I love the finds you post.
thanks Joe Richey
http://www.wholesalerwholesaler.com

It is true that Alexa is unreliable and a lot of their reports have been misleading but there has been numerous instances where Comscore has been unreliable and totally off. Not a complete coverage of the issue though…

I wouldn’t say alexa is “useless.” If you want to know if a domain has significant traffic alexa will still answer the question in the right ballpark.

If someone is claiming their site is in the top 100 sites on the internet and alexa says zero traffic, something is up.

I would have to completely agree. Alexa is so 2001, compete.com is where it’s at…

I like the approach Quantcast.com uses.

And of course, I’m sure Google Analytics could tell its own story, but as usual, they keep it to themselves.

I think before its said to be useless, there is a need to compare more sites than just two.

Darnit, remind me to stop using those Alexa graphs to show how awesome we are.

Both Compete and Nielsen are an order of magnitude smaller than our actual numbers at our web service. Alexa is in the ballpark, with 10% less than actual.

Google Analytics gives the real picture, filtering out bots and stuff that doesn’t count. That’s what we show VC’s and advertisers. That’s what counts.

None of these “guestimation” services can really know, and it’s appalling how OFF Compete and Nielsen are.

Haha, I agree on the Alexa statement. Can’t use the toolbar on my Mac. The headline is so stinging though.

I recall about 5 years ago we were soon to launch a new site, it was live but no one knew about it. About 4 of us had the Alexa toolbar installed while working on the site, and before we told anyone about the site we had a ranking of around 75,000. It still annoys me when people quote Alexa numbers.

Would you, please, please not disclose the flaw in Alexa’s methodology. You will kill my text link pricing model.

Signed-

Text-Link-Ads

Alexa is very “useful” b/c it’s come back from the dead with a vengeance for lay folks who want to quickly diligence a site. Partners, VCs, even customers. It may be way off in many cases, but the fact is, it’s likely being used to diligence you anyway.

Amazon seems to have noticed with FF toolbar etc. Now it’s time to go to the next step and update the methodology.

Mike you are just realizing this now? I have been saying this on CN since it began and for years before that. Agencies knew that if they brought in an Alexa chart, I would end the meeting immediately.

My last post is a bit of goodness - “never bring Alexa to a fight”:
http://www.centernetworks.com/company/alexa

Mike,

Compete aggregates ISP, ASP, opt-in panel, AND toolbar data. Majority of the data is ISP data — natural as most of the volume is there.

For a comparison of Compete vs Alexa vs Comscore vs Hitwise — take a look at:
http://www.compete.com/help#snp2

also Compete, Inc. has been around for 6 years, and raised over $40 million. Hardly a newcomer.

I wouldn’t say Alexa is useless, like everything else you need comparisons between multiple sources. All companies have their way of measuring pageviews and if they all used the same method then you should see the same trend. Alexa is good to see if a site does indeed have traffic. What other free sources are their on the internet?

I still use Alexa when looking at sites though, but def take the results with a grain of salt

I agree in saying Alexa is useless. It’s a tricky business monitoring traffic, and Alexa might have been the way to go 5 years ago, but with such accurate statistics programs out there, Alexa is just about obsolete.

And besides, there’s no way YouTube could be bigger than Google. It probably uses a heck of a lot more bandwidth though…

-Chris
http://www.nerdcouncil.com

Alexa is unreliable. True

Alexa is dangerous as I wrote in reference to the Seattle Startup Index on http://seattle20.sampasite.com.....-truth.htm

But besides that, here is an easy explanation to why Alexa is reporting YouTube having more traffic than Google…. And the story goes like this:

—
A few developers at Alexa are hoping to improve their stats. Lots of people have been complaining about the fact that Alexa doesn’t see AJAX/Flash calls, or that content inside IFrames are not properly accounted for (like in http://www.sampa.com), and so on. They go and they devise a brilliant plan (or, let’s say, some other major stats company did) thinking that PageViews are obsolete. “Yeah, we all know that, so we should throw PV under the bus and start over”, says one happy (yet younger) developer on the Alexa team.

So, the plan is infallible: Lets measure the time people spend on each site and use that as a proxy to our PV stats and that would take care of a lot of problems. They go and they implement this brilliant strategy and now YouTube appears to have 6 times more traffic in PV than it actually has, since people find things quickly on Google, but waste so much time in YouTube watching videos. But the Alexa UI team has been slow to replace the text “Page View” to some new term, like “Page Attention Span”

The End.
—-

Now, this is just a theory. Anyone cares to ping Alexa and figure it out?

-Marcelo Calbucci
http://marcelo.sampasite.com

The

Mike,

Statsaholic has dumped alexa!

http://www.statsaholic.com/about

Ever since we became the #3 site in our category, I’ve been waiting for people to realize that Alexa is worthless

There’s one competitor of ours that has less than 1/4th our members, has been covered here & elsewhere, raised millions of dollars, but has less than 25% of our traffic. At best.

However, Alexa has shown them ahead of us since before they actually launched their website…and to this day, with a million uniques / month, we can’t get ahead of those guys in Alexa. Sure compete & quantcast are light on the numbers, but all the 3rd party data providers are short of Google Analytics or other internal tools.

It’s just embarrassing how far *off* Alexa is…I really, really wish people would stop quoting them.

Alexa was always unreliable because of the type of people who would install it to begin with. But 5 years ago it was reliable and better than nothing, and nothing was all we had when it came to free data until Alexa. That’s no longer true. I’m with Max #6 — I too like the approach Quantcast is using to measure (you put a pixel on your web pages, they track it).

From a results point of view, besides loving all the info and how it’s displayed, Quantcast updates its data on a much more “recent” basis and you don’t get just a “monthly average” but get to see the intra-month results as well. It’s not one data point per month, but many. Given a little more time (it’s still very new) it could be very useful.

that should have said: But 5 years ago Alexa was UNreliable and better than nothing.

Until Alexa ranking is no longer the de facto site ranking number that Joe Schmoe CEO uses for his/her site mental order, I wouldn’t say it is completely worthless.

It just depends on your audience.

I wouldn’t be surprised if YouTube.com DOES have more page views than Google.com.

YouTube only had a .com site until very recently, when it rolled out 9 other TLDs in France, Spain, Brazil, Japan, etc. But those aren’t getting much traction. Basically, everyone in the world who uses YouTube, uses YouTube.com.

Google, however, has had more than a hundred country specific sites for a long time. And people use them. Germans go to Google.de, not Google.com. So, Alexa traffic for Google.com is only a fraction of worldwide traffic to Google.

When was Alexa a reliable tool for measuring traffic? I hate it since the day I came to know that my *private* site was under 1M. Please note that I never leaked the URL of the site. Later I realized that the traffic rank started going up since the day I installed the toolbar. What a waste of time!

I do find it amusing when I see people proudly showing “Alexa” badges on their sites.

Get Rid Of Alexa~~!!!!!!!

Amen, never liked Alexa rankings.

alexa works only with internet explorer - i am thinking alexa users are the same kind of internet users who download crapware and waste their time on youtube.

Hi,

I think the author misunderstands what compete tells. In the link it tells that’s VISITORS, not PAGE VIEWS. And Alexa never tells YouTube has more VISITS than Google.

OK, now lets hope advertisers get it right this time. Stop using alexa to decide were to advertise!

Alexa’s biggest problem is that it is massively biased towards the sort of people who install the Alexa toolbar. Any site which appeals in particular to early adopter types will be have its traffic hugely overestimated.

By looking at the statistics for several websites I run, which appeal to different types of user, I estimate this bias at around 50x. So Alexa’s statistics are worthless for anything more than a basic sniff test. Maybe it has some minimal value for comparing sites which compete directly for the exact same users.

Neither Compete nor QuantCast have this problem with bias. The absolute numbers they provide are far too low, but at least they’re in the right ball park for comparing sites on a relative basis.

Hey! you are in the wrong chart, This is the right chart http://www.alexa.com/data/deta.....google.com that shows the relative position in Alexa Ranking, If you look inside published chart you are making a research about page views!

I believe that Quantcast’s model has it right. Hold the website accountable for reporting traffic, not people that install Alexa.

Michael, you keep making the SAME MISTAKE. It’s embarrassing for me to see you make it so often, but there you go again.

Every time you post about a ratings service you fail to even look at what it’s measuring. Compete (hardly a newcomer, as noted above) measures U.S. TRAFFIC ONLY. ALEXA MEASURES GLOBAL TRAFFIC.

In other words, they can both be right! They are measuring DIFFERENT THINGS.

You don’t disclose what the comScore data is measuring. I’m pretty sure that in your mind, it doesn’t really matter since time and time again you’ve failed to see the difference between US-only, and global numbers.

The United States is certainly very important. But it is only *one* out of *over two hundred* countries on the planet.

Jeesh.

Even i dont think alexa is reliable. Most of the times when ever i received more traffic to my blog that day my alexa ranking would be more than what it would have been on less traffic day.

It seems that Alexa have a time to fight with Statsaholic, but they have not time to improve themselves.

Alexa is very often very wrong.

Quantcast is often very wrong.

Comscore ” ”

Compete ” ”

Even your own Google Analytics is often wrong (due to iframes, buttons on other sites, etc) .

@the person who said GA wont give up their info, I highly doubt yahoo, youtube, etc even use GA. Their traffic would choke GA, and they’d be fools to measure their traffic with a web based app. Thats just a dumb and paranoid comment.

So which is the best? None of them. And its a good thing too. I wouldn’t trust any ratings that come from one source.

Use a scientific approach, like most people do, and gather a little bit from each one and consider how they collect their data. Compile the data and trim here and add there and you’ll have a fairly clear picture.

Alexa is what we call Microsoft, and ComScore, Compete are the “Googles”

Finally I got the proof
Thanks a lot

Alexa measures all regional domains separately for google. These include google.com, .de, .fr, .ro, etc.
For YouTube, as far as I know, there’s only a global .com site. This way the comparison is between US Google users vs. global Youtube users, which might as well be realistic.

Compete is focused on the USA and Alexa offers a Global ranking. That’s the difference. Google has chosen to localize their domains in each country and has many sites with the google.* domain in the top-100. To name a few in the top-50: google.fr, google.co.uk, google.de, google.com.br, google.pl, google.es, google.cn, google.com.mx. For Youtube this localization does not exist and thus includes world wide traffic. I hope you’ll mention this in the post as it explains the difference.
PS: Am I wrong or is/was Compete a sponsor of Techcrunch?

If you want to look at numbers for a very certain demographic which is highly likely to install the toolbar and surf with it, then you can go for it.

But other than that Alexa numbers are like fast food: Easy and they make you believe they are ‘good food’ when in fact they are junk.

If you ‘believe’ in how alexa traffic monitors ‘global’ traffic, get yourself some compareable data. Suggestion for Germany:

IVW are the numbers for done for reach of online and offline media through various mechanisms and the ones I trust (http://ivwonline.de/ausweisung2/suchen2.php).

Sure they do not count ajaxy stuff, but get some numbers from their listing for German sites and compare them of your so called global Alexxa numbers which claims to have a German ‘ranking’ plus to other data and you will see the difference.

You’re comparing Google.com (US) against YouTube.com (worldwide).

The vast majority of Google users outside of the US use Google.co.uk, .de. .fr, .co.in, .com.au etc.

I’m no huge advocate of Alexa, but I think the Alexa stats area accurate in this case - but they don’t mean YouTube is bigger than Google at all - only that it is bigger than Google.COM, which is entirely believable.

After the initial excitement of making it into top 75 very quickly after launching our blog, I realised how skewed the results where.
And over the past year, according to Alexa we lost half our traffic - while we actually grew tenfold…. Compete on the other hand shows us doubling every month - which, alas, isnt true either

For smaller blogs like ours none of the above really shows the full picture. We have about 1000 uniques per day, but 5000 RSS readers per day - and those show up in none of these services. And I have no idea whether that’s 1000 people checking their feed 5 times a day, or 10.000 every other day.
So even I as the sitemanager havent really got a clue how many readers I have exactly. I guess I cant expect it from any outside service…

Compete.com only measures U.S traffic as do Quantcast and Comscore.

In reality, they all suck!

My site has been running for 9 months with good growth in page views / unique visits over that time - yet my Alexa ranking drops further and further. Say no more…

Haven’t many, many other TC readers been trying to tell you for years that Alexa data is useless? — You can throw Comscore in there as well.

Yet, for some reason you kep wanting to use both.

Wake up! Pageview metrics is no more a reliable way to measure web traffic. Instead, I think the next reliable metrics in today’s AJAX era is the average time a user stays with the focus on your page (and browsers could implement this for each individual user, by controlling each TABS, uploading to a central and *OPEN* statistics portal).

More proof that Alexa sucks.

It’s a shame that many web companies like text link ads rely on your Alexa rank.

Alexa has always been pretty useless, but I can attest to it’s ‘value’ in the sale of my previous ecommerce venture. Personally, it’s a fun stat that I always take with a grain of salt. Investors, buyers, and advertisers on the other hand are still obsessive over it.

maybe you are right and they are wrong…moron

digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx
digg suxx

I knew something wasn’t right when Alexa required a toolbar to track site statistics.

Well the problem is that it obviously comes down to how many visitors of the site actually use the alexa toolbar. So for example a technology based website might appear to be more popular simply because more of its users are likely to have the toolbar installed than say a website about dogs.

Yes alexa is biased: for ages they supported only IE!

Lol well maybe they’re basing off of servers, perhaps there comparing youtube.com to google.com (using ip addresses, and perhaps youtubes site resolves to one address, while google is split across it’s many data centers)..still this doesn’t make much sense.

I’m not so sure.

Alexa - i think - ranks it’s data by traffic used. Google is simply a page with text and a few images and javascripts, youtube is mainly videos. Not sure on youtube’s compression scheme, but most the videos must be of a decent size (500kb+). You’d probably need 20-30 page views of google to get one small video on youtube.

A very interesting post!

I think what we are all asking is, how is alexa actually coming up with these figures?

I like your post. I’ve always depended on Alexa rating for my websites but it’s clear from this article that Alexa is completely inaccurate for small or larger sites. What are the major alternatives that are more accurate??

Shafi,
http://www.lxpages.com/

Anyone installed Alexa toolbar? Not for me.

The issue is not how inaccurate alexa is (which is a very boring issue for anyone that deals with internet). the question is why google usage has dropped with alexa users. more non-google search toolbars? yahoo/msn/ask gaining power? My guess - more toolbars and search assistants on an average alexa users pc.

The way Alexa is favoring Youtube its not very far when Youtube will beat Yahoo! in Alexa

Alexa metrics include non-us traffic. I don’t know much about comscore and compete, but I bet they collect U.S. centric data. This may account for the discrepancy. For example, Chinese users would go to google.cn, and maybe youtube.com. This boosts youtube.com’s pageviews, but not google.com’s. There would be a discrepancy if comscore and compete do not measure international traffic.

See the percentage of users from the U.S. for youtube.com and google.com:
http://alexa.com/data/details/.....outube.com
http://alexa.com/data/details/.....google.com

Google get’s 25% US traffic, YouTube gets 12.5%. These figures support the above claim.

what’s worse is how many people think Alexa is accurate, particularly in advertising and marketing. I can’t stand sites like it. I think they should be shut down because they’re creating a lot of false hype and inaccuracy in the market, without users having a concrete understanding of how it all works.

Just my .2. Analytics are obviously a sore spot. lol.

What a worth while post. Thanks for all the facts and your counter-arguments, proving without a doubt that you’re are correct in your theory. Its just so obvious, youtube greater than google? UNPOSSIBLE!

I mean how could we even think for a second that there could be a site more popular than google? That is total and utter blashmephy there. I mean I don’t know a single person who has even heard of youtube, or anybody who browses around there for hours. I often find myself getting lost in google results. I often just search for things just to search and I never leave that page. Its just so full of things to waste my time with.

Thanks for once again confirming my beliefs that there is no fan boyism on the internets. And we can look at the tubes from up above and realize that google is in all of them.

@ warren, you sound like the guy that thought paddleboats would never be replaced by steamships.

never say never. i’m sure there could easily be a day when somebody else is larger than Google.

Would everyone just click on my site above. You don’t have to oder or anything, I just want to double my counter and then get Google to refund my ad words! Thanks folks!

Mike,
Anyone with any research knowledge has known this for years. The first questions you should ask of all research are “where do these numbers come from?” and then “Is this representative of the population?” Companies like Hitwise, Comscore, NetRatings, and even Compete go to great lengths to normalize their data and make sure it is statistically representative of the online population. That’s why their services are not (for the most part) free. Any data source (free or not) should always be questioned.

Your tendency to sensationalize questionable data undermines your credibility, and this post further reveals your analytical skills to be second-rate. In order to make your writing more credible and informed, I recommend that you gain an understanding of data collection methodologies by spending some serious time with people from the measurement companies.

When I met you over a year ago in reference to this post http://www.techcrunch.com/2006.....echnorati/ and told you that Alexa was useless, you did not believe me. I’m glad you finally saw the light.

Alexa is pretty accurate for sites with

I think there’s a misunderstanding here. Comscore combines all the local google domains, and alexa compared google.com and youtube.com

If you take a look at the alexa 100 you’ll see that there are >25 local google domains listed there. If you combine the data of all these domains you’ll get a difference similar to what Comscore reports.

Not that inaccurate after all

I wonder if Alexa’s numbers are skewed by embedded YouTube videos. That gives YouTube tremendous reach, but it’s not the same as surfing directly on YouTube.

Garbage in… Garbage out…

Ed Kohler,

Alexa’s numbers are not skewed by embedded YouTube videos. This is because the toolbar don’t make requests to Alexa for embedded media, AFAIK.

For example, check out the firefox toolbar (http://www.alexa.com/site/download/). When I a website I’ll see the traffic graph for that website. I won’t see the traffic graph for an embedded video. This can be verified with a tool like tcpdump.

Ernesto,

my point exactly.

Again…

How many times do we have to say that Alexa data is fucking pointless? We’ve been telling him that for years.? Why the fuck is this news?? Of course it is fucking useless. It isn’t a measure of real internet traffic…. is a measure how many idiots using the stupid Alexa toolbar visit a fucking site.

I wonder if Alexa gets fooled by advertisements on sites which make multiple requests, or if the amount of bandwidth served is affecting results also.

I have a rather large male appendage. I wonder if Alexa could measure THAT accurately?

@LeeAnn Prescott

I have to agree with you. Anyone with one iota of technical competence would suspect Alexa sucks the moment they learn how the data are collected (self -selected toolbars.)

Anyone who owns two blogs in different niches has the data to discover the relative error in traffic detection is 4 orders of magnitude. My very simple comparison of a knitting blog to a “blog about blogging” blog suggest that on average Alexa is 20,000 times more likely to count an honest to goodness visit to a “blog about blogging” than an honest to goodness hit at a knitting blog. Discussion in Alexa: Only off by a factor of 20,000!?

With errors this large, it’s hard to believe anyone with IT credentials would only now be discovering Alexa is useless.

Regardless of these numbers, I find it fascinating that a company that has existed less then a year and a half has sustained this amount of incredible growth… just makes you realize how precarious “global web positioning” really is if an unknown can possibly become bigger then the worlds most successful online conglomerate!

Jon

Alexa is dead.

Well Alexa’s ranking as stated by alexa itself is biased. It is based on a small percentage of a whole lot of internet surfers. But still alot of value is given to Alexa’s ranking by some VC’s and investors.

I will keep alexa because when i had problems with internet exployer and could not sighn on i used alexa sighned on and then sorted out my problems with internet exploer.Alexa was a godsent to me

Amazing, although the problem is what John stated above, they are only based on a small percentage of internet users that happen to have downloaded the toolbar.

Alexa doesnt proof anything so far but just simply put in ranking for website with uncertain data.

When a ‘tool’ show an un-natural number of SEO and web marketing related websites in it’s top lists, it shows that the tool is biased to only a certain specific user group.

That being the group that are most interested in web traffic and the group that are most likely to download the toolbar…

It’s not a true reflection of a site’s visitor numbers or web traffic at all.

I’d bet the SAGA website isn’t on the list, but any crinkly who surfs the web will definitely go to the SAGA website. (How else does one obtain Rolling Stones concert tickets these days?) Why wouldn’t SAGA be listed? Because the crinklies probably won’t have downloaded the Alexa tool bar.

Alexa has been nothing but a bit of fun for a number of years. Not worth the ‘paper’ that it’s written on. As one loo roll manufacturer used to put it in their telly adverts.

‘It’s soft, strong and thoroughly absorbant’

Nice article, but this is nothing new.

Alexa is going down!

I like Alexa - visited Compete and it seems that it only works for the big guys… and you have to jion, how interesting….when do they start charging a fee????

http://omadsense.com

1. Alexa might be right with all the linking people do for YouTube and with all the redirected traffic.

2. Many people started to use the Internet only for YouTube. Many companies report that their employees are addicted to the website.

3. The mainstream advertising (with the elections campaign) also influenced this rise.

Great post- I’ve heard of comscore, hadn’t heard of compete- I’ll look into those now, thanks!

Brian

This single mistake by Alexa might end up costing them quite a lot. This article, 20 tracebacks, tracebacks to those blogs accounts to a lot of readers, and possibly people in the business of using these services.

Maybe we are comparing apples and oranges here, Michael.

Youtube is very popular worldwide and its popularity is increasing with an acceleration, and it has a single brand and domain for its worldwide operation.

But Google operates with 157 different domains in the world. Google mostly redirects people to local web sites in their territories. All local google sites are listed seperately on Alexa: Google.com is #3, Google.FR is #24, BR is 25, UK is 27, and the list goes on. The Alexa Global 500 is full of Google web sites.

It is not surprising that the pageviews per session average is better for Youtube, twice for a comparison between Google.com and Youtube.com, and mostly the same for international Google versions vs Youtube.

All these combined, and many other parameters which would be easily added into the equation, the resulting data would show that Youtube has been a bigger source of traffic than Google.com.

AdditSome simple maths would help to examine the big picture.

liked the last comment of Devrim

Hey,
recently I did a research on Most of the PR10 sites (Page Rank 10)..and compare them against Alexa and Compete. results are somewhat strange.
http://www.vebguru.com/?p=86
-Nish

Hi,
I was using Alexa stats earlier. So i have to find a better alternative. Thanks Michael.

I have to agree with the author. I think since Amzon bought Alexa it has steadly become irelevent and incorrect in measuring traffic.

Everybody knows that Alexa is not accurate. But this doesn’t mean it has no value. It is still the easiest way to have a first picture of a site’s traffic.

Look we all are talking about Alexa … why? Because there is simply no other alternative. With many I agree, installing alexa toolbar increases traffic rank. It`s not fair, but it`s like this. Google analytics, any log analyzer programs or your own calculation nothing are correct. It`s a combination of all!!

All SEO Tools, Link poularity checkers, search engine saturation checkers, and yes, the SEO guru’s bible, wordtracker also are useless by your standards.

Overture doesn’t give great results either.

No tool can accurately guage everything. You have to use them all and get an idea of how well you are doing and not believe you have a completely accurate assessment at any of them.

Depending on tools and not your brain is not a good way to run a business.

Also, remind me again why we care to check google or youtube’s alexa ranking? Unfortunately I don’t own either of them nor are they my clients, so pretty much a “too much time on your hands” issue here?

… and what other services believe that there are any other countries in the world other than US and measure non-US traffic??? :)))))
funny stuff….

hi i enjoyed the read

Who cares what Alexa say’s, it’s accurate anyway!

Its not accurate, damn, there is no “Edit” here……

Hi Micheal,

You have got this badly wrong and here is why:

http://www.sitelogic.co.uk/tec.....not-alexa/

Sorry about the post title, I may tone that down a bit

The problem with Comscore is similar to Neilsen in that there are only a limited number of households being tracked to determine or extrapolate the supposedly viewing habits of the Internet as a whole. All these metrics have major limitations in their ability to track accurately what is going on across the Internet. You have to take them with a big grain of salt.

In the end they are all a means of getting some idea of where people may or may not be spending their time online. Only the log files at a given website can truely tell you anything meaningful, and even then, trying to dicipher true unique visitors is black art at best.