21 replies to this topic

[JohnMu]

How many of you use Google to check the spelling of a word? I know I do that, sometimes. When you search for something like mispell that is obviously misspelled, it automatically suggests a new query:

Did you mean: misspell

Google
Looking at my logs, I noticed that an average of 5 to 6% of all visitors from Google used the spell-correction to make it to my site (this is the same across several sites, the rest of the numbers are from one site only for the last 6 months). I was really surprised at that number!

7.7% of my Google visitors came in with the Firefox Google-search box (which provides income to Mozilla), 2.2% came through some other custom search engine / Adsense for search, 1.0% came in from a Dell branded Google search. 0.4% had their Google display 100 entries/page, whoda thunk. 3.5% had safe-search active (at least in the referrer URL, in various forms), 1.2% had it turned off (in the URL).

Yahoo!
Of all my Yahoo users, only 1.8% used the spelling correction (1.7% used the spelling suggestion on top of the results, 0.1% used the one on the bottom) -- which could mean that either Yahoo users know how to spell or Yahoo doesn't correct misspellings that well, or perhaps even, that Yahoo sends users to the proper search results even if the query is misspelled. I wonder which one it is?

Yahoo displays a lot of related queries: over 2.5% of my visitors (from Yahoo) used a related query suggestion to make it to my site. The first related suggestion got 1.7%, the second one 0.6%, so the top suggestions seem to be useful (to my visitors at least).

MSN/Live won't tell me much and the 3 visitors I got from ask.com aren't really statistically relevant.

Oh and "I'm feeling lucky"? Only 0.2% of my Google-visitors used that. Maybe I just need more #1 rankings :naughty:

Check your sites!
Go ahead and try it with your own log files! I put together a small tool to pull those numbers out: http://oyoy.eu/site/sereferrer/

It has an online version (can only process log files of a limited size) and a Windows-PC based version (which I used to check 10GB of logs - finally something to do with all those old files).

Let me know what you think! Oh, and please post your numbers, it would be great to compare!

John

[eKstreme]

Nice one John! Very nice. Thanks!

I won't share stats about ekstreme.com because my sorry host has lost my log files. But what little I have gave some interesting results

Pierre

[rynert]

Thought I would have a play, but daily logs are over 150MB - tried the download version but it crapped out, complaining of too many files... what is the limit?

[rynert]

Thought I would have a play, but daily logs are over 150MB - tried the download version but it crapped out, complaining of too many files... what is the limit?

Just ran if for 7 days. took a little while, then once finished I hit 'RUN' again cos I thought that was how I view the results! lol DOH!

Anyway.. running again now....

[rynert]

HTML-View (public view)

Gives page cannot be found...

But anyway, based on around 25k searches

Google
6.7% misspelled
0.4% I'm feeling lucky

Yahoo
5.6% From related (rs=0)
1.0% Misspelled (SpellState)


In all cases between 85-90% were from Ranking 1-11; acts as a reminder than page 2 is not much good in this game!

Interesting was the word length of the searches;

Google
36.8% Two words
31.1% Three words
26.6% 4-9 words

Yahoo
61.1% Three words
19.7% 4-9 words
18.6% Two words

MSN
38.7% 4-9 words
26.3% Three words
25.4% Two words

It varies quite a lot between the 3 sites - looking at my own logs it is probably down to lack of volume and therefore long-tail searches, Yahoo is 20% 'planning a wedding' which distorts it I guess.

Interesting stuff - thanks

[JohnMu]

Cool - glad to see that it's working !

Those word lengths are interesting... I wonder if that is because people search for more words on MSN/Yahoo or if the shorter phrases are much more competitive?

John

[EGOL]

I have one site that sells a product that has a formal name, an informal name, a slang name, and an ignorant usage. When I started the site a few years ago, google did not distinguish them and I got that traffic through optimization. Now, all pages - no matter how they are optimized show in any of those searches - with or without the name variant.

[SEOigloo]

And yet, click through rates, link popularity, etc. may be at play in this, because Google is no longer auto-correcting the word Flickr when you search for it....

Miriam

[bragadocchio]

Nice tool!

Those word lengths are interesting... I wonder if that is because people search for more words on MSN/Yahoo or if the shorter phrases are much more competitive?

I have the opposite result, with longer queries finding results on my pages through Google than Yahoo.

Google

4-9 words 29.1%
Two words 19.3%
Three words 17.9%
Single word 3.8%
10 or more words 1.0%

Yahoo

Two words 38.4%
Three words 25.9%
4-9 words 25.6%
Single word 7.8%
10 or more words 1.4%

MSN
Three words 34.2%
4-9 words 29.9%
Two words 26.1%
Single word 8.3%
10 or more words 0.9%

Misspellings in Googe: 5.3%
Misspellings in Yahoo: 1.7%


Ranking 1-11 Google: 86.0%
Ranking 1-11 Yahoo: 84.2%
Ranking 1-11 MSN 79.3%

And yet, click through rates, link popularity, etc. may be at play in this, because Google is no longer auto-correcting the word Flickr when you search for it....

Google is showing me a mid-page query refinement pointing to Flickr when I search for flicker.

[Mano70]

Really nice tool. I did run Strictly CSS quickly through the tool (not a complete run of all my log files):

Google:
4-9 words 42,5%
Three words 30,6%
Two words 17,5%
Single word 3,2%
10 or more words 0,1%

Yahoo:
4-9 words 38,7%
Three words 36,0%
Two words 21,5%
Single word 2,3%
10 or more words 1,1%

MSN:
4-9 words 34,1%
Single word 27,3%
Three words 22,7%
Two words 15,9%

Misspelled Google: 4,3%
Misspelled Yahoo: 0,8%

Ranking 1-11 Google: 69,0%
Ranking 1-11 Yahoo: 69,0%

I wouldn't put too much into the numbers from Yahoo or MSN, the amount of visitors from them are sooooo small compared to Google.

Edited by Mano70, 30 June 2007 - 06:52 PM.

[bobbb]

Tested 1 month on two french language sites (nothing with as much traffic as most here are used to getting)
M and Y! are negligable

Google site 1

Misspelled 4.6%

Three words 34.3%
Two words 27.8%
4-9 words 22.7%
Single word 2.9%

Ranking
1 - 11 87.7%
2 - 12 2.9%
11 - 21 2.0%

Google site 2
Misspelled 9.8%

4-9 words 40.7%
Three words 32.5%
Two words 22.4%
Single word 3.6%

Ranking
1 - 11 78.1%
11 - 21 9.3%
41 - 51 4.1%
51 - 61 2.3%
21 - 31 1.8%

Edited by bobbb, 02 July 2007 - 03:41 PM.

[JohnMu]

It's interesting that the percentage of misspelled queries is about the same on your sites .

I wonder ...

How many visitors are going to a different site instead of accepting the spelling suggestion? Would the majority choose a site in the listing or would they choose the spelling suggestion and go from there? Would it make sense to target misspelled phrases if Google is already suggesting a correction (of course, assuming that Google is the main source of your traffic)?

In other words - is optimizing for misspellings now a waste of time?

John

[mikeg]

Hi I am kind of new to these forums and have been reading this particular thread. I thought I would join in.

"In other words - is optimizing for misspellings now a waste of time?"

In Google; no its not a waste of time. At least for me it is not. I target trademarked keywords that are commonly misspelled on an organic level to supplement the cost for placing an ad there. It has traditionally converted many sales regardless if there is a "did you mean keyword" link there.

One thing that I have noticed over the past 2 business quarters is that typo keyword in yahoo are nowhere to be found. I used to rank for them and they are no longer there. However the PPC ad is still present. It makes me think that Yahoo is actually cranking down harder on typo keywords more so than google. I know this is not about Yahoo and thats nother story all together, but I do quite well organically in Google for typos and it does convert and has done so for several years now.

[redcardinal]

Just grabbing now. Sounds very interesting.

But come on, seriously, where do you find all the time John?

[JohnMu]

Good to see you here, Red

My day has 32 hours.

I get up in the morning and cycle close to the speed of light down the hill near my place - it turns back my relative time by about 3-4 hours (depending on my daily form). I do the same on the way home.

Mike (welcome to cre8asite forums :wave: ), what percent of your traffic comes from misspellings? (non-corrected) Is it more than the 5-6% that seems average?

John

[mikeg]

5-6% sounds about right...makes me think that in spite of the spelling that is what they were looking for in the first place. 5-6% is not a lot, but when you get a good mispelled keyword that converts to a sale, its worth it. At least thats how it seems to be for me on a historical plain. :naughty:

[Black_Knight]

cycle close to the speed of light down the hill near my place ... I do the same on the way home.

If only we all lived where you can cycle downhill both ways.

[JohnMu]

4 dimensions can be awfully twisted when you're going that fast, Ammon

I just stumbled into this -- http://www.google.co...bs/britney.html

The data below shows some of the misspellings detected by our spelling correction system for the query [ britney spears ], and the count of how many different users spelled her name that way. Each of these variations was entered by at least two different unique users within a three month period, and was corrected to [ britney spears ] by our spelling correction system (data for the correctly spelled query is shown for comparison).

I bet they have some interesting data on this topic...

John

[redcardinal]

I get up in the morning and cycle close to the speed of light down the hill near my place - it turns back my relative time by about 3-4 hours (depending on my daily form). I do the same on the way home.

Normally I'd think you were pulling my leg, but seen as it's you....

[Respree]

Interesting how some advertisers are paying for terms like 'wesbite', trying to traffic to their wesbite, er, I mean website.

http://search.yahoo....a...ss&ei=UTF-8