Of course I'm not the first person to point out how off-kilter web measurements are.
A modest proposal
Let's look at a search referral traffic the way a site owner would.
I picked a basket of medium-to-large websites and looked at the inbound search traffic percentages using Hitwise. I included Topix in this mix, both because it's a representative content site, and also because I could double-check the Hitwise numbers against our own server logs and 3rd party measurements from Google Analytics. As it turns out, the relative inbound referral ratios agreed between Hitwise, Google Analytics and our own server stats.
The results:
| Site | hitwise-google | hitwise-yahoo | hitwise-msn | hitwise-ask | Yahoo | MSN | Ask | hitwise-total | |
| apple.com | 8.62 | 2.38 | 1.69 | 0 | 67.9% | 18.8% | 13.3% | 0.0% | 12.69 |
| craigslist | 7.48 | 3.4 | 1.17 | 0.12 | 61.5% | 27.9% | 9.6% | 1.0% | 12.17 |
| ebay | 10.12 | 3.36 | 2.57 | 0.44 | 61.4% | 20.4% | 15.6% | 2.7% | 16.49 |
| flickr | 17.72 | 7.26 | 1.34 | 0.45 | 66.2% | 27.1% | 5.0% | 1.7% | 26.77 |
| nytimes | 16.67 | 2.84 | 1.34 | 0.53 | 78.0% | 13.3% | 6.3% | 2.5% | 21.38 |
| topix.net | 40.5 | 10.02 | 0.65 | 1.56 | 76.8% | 19.0% | 1.2% | 3.0% | 52.73 |
| tripadvisor | 47.57 | 5.87 | 3.51 | 1.42 | 81.5% | 10.1% | 6.0% | 2.4% | 58.37 |
| usatoday | 6.43 | 2.07 | 1.4 | 0 | 64.9% | 20.9% | 14.1% | 0.0% | 9.9 |
| wikipedia | 48.36 | 10.98 | 3.66 | 2.57 | 73.8% | 16.7% | 5.6% | 3.9% | 65.57 |
| youtube | 12.97 | 2.28 | 2.16 | 0 | 74.5% | 13.1% | 12.4% | 0.0% | 17.41 |
| Average | 70.6% | 18.7% | 8.9% | 1.7% | |||||
What I did
I did a simple average of the percentages instead of a weighted average, to offset the chance that a particular site was being unduly favored by a particular engine. (It doesn't look like that is happening though; both Yahoo and Google favor Wikipedia and IMDB in their top organic outbound referrals, so they seem to be sending traffic to the same kinds of places in their listings). These numbers probably undercount Ask, because they were below the top inbound referrer cutoff for some of these sites.
I'm not a professional analyst, and my approach here is pretty back-of-the-napkin. Still, it confirms what those of us in the search industry have known for a long time.
The New York Times, for instance, gets nearly 6X as much traffic from Google as it does from Yahoo. Tripadvisor gets 8X as much traffic from Google vs. Yahoo.
Even Yahoo's own sites are no different. While it receives a greater fraction of Yahoo search traffic than average, Yahoo's own flickr service gets 2.4 times as much traffic from Google as it does from Yahoo.
My favorite example (not included in the above stats): According to Hitwise, Yahoo blogger Jeremy Zawodny gets 92%
of his inbound search traffic from Google, and only 2.7% from Yahoo. :-)
"We see little to stop Google from reaching 70 percent market share eventually; the question, really, comes down to, 'How long could it take?" -- RBC Capital Markets analyst Jordan Rohan.
Welcome to the future, we're already there. To paraphrase an old industry saying about IBM...
Google's not the competition, Google's the environment.
Comments (28)
Interesting.
That number is pretty accurate for my site. I wonder what their methodology is...
Posted by Jeremy Zawodny | December 19, 2006 10:36 PM
Posted on December 19, 2006 22:36
I have another theory - Yahoo doesnt provide good search results, and hence user has to search again on yahoo. Therefore yahoo has more "search query" traffic. But since user doesnt click on the result that often, the referrer traffic (our of yahoo) is lower. I think this methodology, of measuring the referrer traffic, correctly reflects the "search" traffic. Yahoo has artifically higher search page views because of their not so good results :)
Posted by Analyzer | December 20, 2006 12:27 AM
Posted on December 20, 2006 00:27
It is conceivable that the more techie or academic a site's audience is, the more it is apt to attract Google users.
However, looking at the 2006 TOP search terms from the major Search Engines - there is an entirely different audience out there that appearently would be happy using just any search engines because for these simple terms, all SERPs would all bring relevant results
seoptimization.blog.com/1374224/
BTW:
How does one gain access to those stats for the various large Website's used as examples in this post
Posted by Search Engines Web | December 20, 2006 12:53 AM
Posted on December 20, 2006 00:53
You need a subscription to Hitwise. It's expensive for an individual but an absolutely essential tool if you're in the online industry. e.g. see http://www.seomoz.org/blogdetail.php?ID=1582
Posted by Rich Skrenta | December 20, 2006 5:54 AM
Posted on December 20, 2006 05:54
Rich, I think the difference is "search referrals" versus "total searches performed". They are different, but that doesn't explain the relative differences in market share.
My own blog traffic breaks down like this; 69.5% from Google, 1% Yahoo, 1% MSN, 10% links from other bloggers, and 18% links from my own blog to other posts.
My guess is you are all right (You, Hitwise, ComScore)but the definitions and parameters are different. I wrote a blog today on my findings http://dondodge.typepad.com/the_next_big_thing/2006/12/googles_true_ma.html
Don Dodge
Posted by Don Dodge | December 20, 2006 7:31 AM
Posted on December 20, 2006 07:31
Don,
WebSideStory does claim to measure searches referred. Linked under 'suffer' above, or here -- http://www.websidestory.com/company/news-events/press-releases/2004-03-30.html
Posted by Rich Skrenta | December 20, 2006 9:46 AM
Posted on December 20, 2006 09:46
Rich, that link to WebSideStory is from March 2004...almost 3 years ago. Google has gained significant market share since then.
Here is a link to a more recent WebSideStory press release that says Google accounts for 75% of referrals in the UK. http://www.websidestory.com/company/news-events/press-releases/2006-03-24.html
I checked their site and couldn't find any more recent press release.
Don
Posted by Don Dodge | December 20, 2006 12:57 PM
Posted on December 20, 2006 12:57
Ah, good point Don. I stand corrected. Thanks for the more up-to-date link.
Posted by Rich Skrenta | December 20, 2006 2:36 PM
Posted on December 20, 2006 14:36
We seem to get ignored, but I will try again :)
We run MapStats, which is a stat program only available to bloggers. We track roughly 350-400k unique visitors per day, so the sample size is pretty large.
For this month: http://mapstats.blogflux.com/showstats.html - 88% of SE referrals were Google. Compare that with our first month - http://mapstats.blogflux.com/showstats-2005-10.html
The others are not even 'also-rans' ... they are more like 'who?'
Posted by AhmedF | December 20, 2006 7:36 PM
Posted on December 20, 2006 19:36
Have you heard about the long tail? Could it be that the smaller sites reflect the ComScore/Nielson scores better?
Posted by Pras | December 21, 2006 10:43 AM
Posted on December 21, 2006 10:43
Authority. Look at the sites you chose.
Posted by nuevojefe | December 21, 2006 3:18 PM
Posted on December 21, 2006 15:18
Google has a bias toward over-representing content on authoritative domains as being relevant in their search results, thus numbers from these types of sites will skew high toward Google. In that regard, Google is pushing out the smaller players.
Also since MSN is so easy to manipulate lots of smaller players will be overrepresented in their results.
Posted by aaron wall | December 21, 2006 9:13 PM
Posted on December 21, 2006 21:13
Aaron -- interesting. Are you saying that, for instance, these sites which are under-represented in google (but are still ranked to some extent) are getting different fractions of traffic from G/Y/M/A? I'd be curious to see...
Andrew Goodman has pointed me at the Enquisite data, which seems to represent more of the smaller sites than Hitwise.
Posted by Rich Skrenta | December 21, 2006 9:22 PM
Posted on December 21, 2006 21:22
Interesting Insights, but the thing is every body out there, who does even a little SEO for their sites, are all Optimizing for Google, when they do optimize for search engines.
Publishers love it, Google loves it.
Posted by Tony | December 22, 2006 2:42 AM
Posted on December 22, 2006 02:42
Network effect...
Posted by Rich Skrenta | December 22, 2006 8:18 AM
Posted on December 22, 2006 08:18
Hi Rich
Yeah. It is hard to come up with exact percentages for anything that represents smaller sites beyond a site by site basis though. In some instances I have MSN sending me far more traffic than Yahoo!, but generally even with those Google still sends far more traffic than the others unless my sites are banned in Google. Also worth noting that Yahoo! seems to be moving more toward a Google-like authoritative domain bias, but even with that they have crawling problems with some of the larger sites.
Posted by aaron wall | December 22, 2006 10:12 AM
Posted on December 22, 2006 10:12
So, the big question will be, will the differences between searches/Comscore and referrals/Hitwise narrow once Panama launches and has a few quarters to optimize? It makes sense that Google's relevance-driven engine in fact provides a higher ratio of click-thru's to searches (or, a higher referral share than search share). I'll be interested in seeing this data next fall...
Posted by Bill Lockhart | December 27, 2006 12:48 PM
Posted on December 27, 2006 12:48
As a follow-up, anyone have this type of data from a year ago? Has Google's lead expanded, and is this evidence that it serves up increasingly relevant results (and also that their marketplace is getting larger?) Thanks.
Posted by Bill Lockhart | December 27, 2006 12:54 PM
Posted on December 27, 2006 12:54
i need the algorithmes of the search engine and theme o()for comparison.
thanks alot
Posted by fatemeh banar | December 29, 2006 11:14 PM
Posted on December 29, 2006 23:14
It was refreshing to see someone confirm a pattern I've often wondered about -- the actual share of new visitors coming to our sites from Google is far greater than their purported 50% market share.
I've always wondered why the "conventional wisdom" about Search Engine market shares was so inconsistent with the data for our sites.
Based on my own experience, I don't think "small sites" or "long tail" sites -- per se -- get more than 20 - 30% of their search traffic from Y, MSN and Ask.
However, it's possible there are specific classes of sites that are receiving a large share of traffic from the smaller SE's. For instance, there are undoubtedly some small sites that are SEO'd specifically for Y or MSN, and there may be lots of "MFA" sites (sites with little or no real content, filled with text ads that receive high click throughs) which get filtered out of Google, but slip past the filters at MSN and Y.
Posted by Ben Johnson | January 1, 2007 6:24 AM
Posted on January 1, 2007 06:24
Most of the top sites are paying big dollars to get extra traffic from search engines using paid search results, whereas the littler guys aren't doing that as much. Given Google has the best clickthru rates in the industry for paid ads, I think it's possible the results are skewed a bit in Google's favor. Perhaps this explains the difference between HitWise and Google Analytics results.
Also, I trust Google is in fact not evil, and mostly impartial, but if there's any doubt in a Google Analytics formula, it would be in Google's interest to skew that data in favor of Google, because it would inspire website owners to optimize on Google for ads, SEO, etc....
Just some things to consider...
Posted by Jack Pitts | January 24, 2007 8:19 AM
Posted on January 24, 2007 08:19
Thanks for the analysis by the way, great work.
I'd like to see it expand beyond just the top 100.
Also, I'd like to see the results you presented in the chart above for future months, every month. Would be a great source of info and competitive with Hitwise/ComScore/Nielsen in a certain respect-- probably great for garnering extra traffic every month (and thus revenues).
thanks.
Posted by Jack Pitts | January 24, 2007 8:22 AM
Posted on January 24, 2007 08:22
Sildenafil citrate, sold under the name Viagra is the first of a new group of medication which allows adequate sexual stimulation, relaxes the blood vessels of the penis and helps erectionbuy cheap VIAGRA online
Posted by jessicaholin | January 30, 2007 1:40 PM
Posted on January 30, 2007 13:40
I get the vast majority of my traffic from Google, but have noticed its also aged related. Sites that focus on older markets do tend to get more yahoo users.
Posted by SizeGenetics | July 16, 2007 6:03 AM
Posted on July 16, 2007 06:03
We don't know the main logic behind google and yahoo or other search engine .All are going through are own search theory.In coming year we will know who will win the race .Goole also launch croma browser this is also not a good news for other browser.Hope this will capture market soon.
Posted by Market Watch | March 12, 2009 8:57 PM
Posted on March 12, 2009 20:57
Yahoo is also great web site with a huge traffic but their search engine not covers most of the sites. thats why google is best yahoo is good with its email service
Posted by Asim | March 29, 2009 11:20 AM
Posted on March 29, 2009 11:20
If I compare the baclinks, inlinks etc. in Google, Yahoo, MSN Live, DMOZ and Alexa etc. I get such incongruent results it's unbelievable. If I go to META search engines it gets even more confusing. If I use quotes around the domain, like "www.viaregia.ch" instead just links:www.viaregia.ch I again get different results, varying wildly from 130 to 21,000 ... (with other sites taken at random, but FOR THE SAME site, only difference the quotes, although they should not matter.
Posted by Franz | May 1, 2009 2:55 PM
Posted on May 1, 2009 14:55
i get a mishmash of results around now that you guys mention it, google pulls in a lot more age related ones which is concurren t with what you said.
Posted by Daniel | July 1, 2010 11:53 AM
Posted on July 1, 2010 11:53