54 replies to this topic

[Randy]

Not to mention a very real issue all search engines struggle with, namely trying to determine what someone actually wants to see based upon the search term they used.

As a quick and easy for instance, if a user was searching for "apple" are they looking for the computer company, the information about apples, places to buy apples, pictures of apples, something about New York city (the Big Apple), or something else entirely.

The moral of the story being that depending upon exactly what that user was looking for several of the top ranking sites for the term may be very, very UNrelevant. But to another user who inputs the same term may find one of these UNrelevant sites to be exactly what they're looking for!

It would be folly for them to put too much weight on clickthru's or even time on site stats, number of pages viewed (assuming they have access to all of the data) without first being relatively certain what a person is looking for when they conduct a search. Something none of the engines have gotten a good handle on as yet. Boosting one site up or knocking another down without first knowing why someone is searching for what they've input as a search term is just asking for trouble, and would be totally wrong as often as it would be right.

[SEOOZ]

The folks at Google are so concerned with their results being artificially manipulated, I find it difficult to believe they'd leave such a honkin' big -- and OBVIOUS -- loophole wide open and inviting abuse. I mean, if a humble webmaster like me could come up with this plan in a matter of minutes, just think what some experienced blackhat with some mad programming skilz and a bit of time on his/her hands could devise.

--Torka

Google are pretty good a detecting pay per click fraud these days (unlike Yahoo). Dont you think that they would use the same technology to detect fraud in organic results?

[SEOOZ]

scouseflip,

I have had some SEO marketers tell me that Google Analytics can be used as advantageous to your ranking. Because they have all your stats to see what trends to use. How does anyone you feel about this?

Its possible that they use analytics data. Who knows what they use it for.

[Zishan Ahmed]

As a new member of the forum, i have really learned and enjoyed going through these thread...truly it was a good learning experience...

[scouseflip]

Google are pretty good a detecting pay per click fraud these days (unlike Yahoo). Dont you think that they would use the same technology to detect fraud in organic results?

How good are they really at detecting click fraud? I have never dealt with Google directly on this matter so would need input from others to qualify it, but I have heard a number of people say (mostly in reference to Adsense to be fair) that Google do very little to monitor or respond to click fraud, and that actually they just respond to claims made by advertisers - often refunding the disputed clicks without any evidence that the claims have been looked into.

I'm sure they can easily detect very obvious fraud - a 200% increase in clicks with one IP accounting for the extra for example would be quite easy to spot. But as Torka said, it does not look very secure from abuse. Even as a learner I can easily fool Analytics into recording a page visit from my own computer even though the IP is blocked from Analytics.

I don't really like the idea that Google has access to all our stats through GoAn but it was always going to be likely that they would use this data somehow and I accepted that - how else am I going to get such a nice analytics package without having to put hand to pocket? But I really hope they are not going to use it for their algo, at least not as a big part of it - as much as I respect that they are not daft, I just can't see how they can be sure that it won't be abused.

[gangwisch]

How good are they really at detecting click fraud? I have never dealt with Google directly on this matter so would need input from others to qualify it, but I have heard a number of people say (mostly in reference to Adsense to be fair) that Google do very little to monitor or respond to click fraud, and that actually they just respond to claims made by advertisers - often refunding the disputed clicks without any evidence that the claims have been looked into.

I'm sure they can easily detect very obvious fraud - a 200% increase in clicks with one IP accounting for the extra for example would be quite easy to spot. But as Torka said, it does not look very secure from abuse. Even as a learner I can easily fool Analytics into recording a page visit from my own computer even though the IP is blocked from Analytics.

I don't really like the idea that Google has access to all our stats through GoAn but it was always going to be likely that they would use this data somehow and I accepted that - how else am I going to get such a nice analytics package without having to put hand to pocket? But I really hope they are not going to use it for their algo, at least not as a big part of it - as much as I respect that they are not daft, I just can't see how they can be sure that it won't be abused.

scouseflip,

I feel like you have a negative attitude towards Google Analytics, which is ok. As you said it's the price of having a free tool. But do you feel at all that this could be advantageous to us in anyway, or am i thinking out of the box too much.

[s2b]

There is nothing like clicking fraud except in case of clicking advertisements. As many time you click a link from results, it adds to its total page views, which is, however, different from unique page views. And this practice is good for nothing unless u want to do some fingertip exercise.

[madams]

Clickthroughs alone would not be a good measurement, however, I believe they also attempt to measure time on site too. So if you come right back to the SERPs and click to other sites, it shows the first wasn't relevant to the search query (at least for that particular user). If you don't come back to the SERP, then presumably, the searcher found what they were looking for.

How do you qualify this statement?

I am looking at a site and I go to make a cup of tea and and cheese and cucumber sandwich (I´m English)

The site I was looking at has me logged for 25 minutes. Does this make the site relevant to what I was searching for?

I think not.

Another exampl: I am looking at a site and it seems to have what I want.

I go through 20 pages and I still cant find what I´m looking for!

In the end I go back to search to find it.

I find what i´m looking for on the first page of a new site.

Does this mean the first site I looked at is the best?

I´m going to bed....

[gangwisch]

No, it means that if you search for something click on a link and come back in 5 seconds, they did not return a good website for that search term

[projectphp]

Google are pretty good a detecting pay per click fraud these days (unlike Yahoo). Dont you think that they would use the same technology to detect fraud in organic results?

How can it be fraud? Whose paying/losing?

Most of the sites i optimise rank 1st for highly competitive keywords using strategies based on the above so i think that i know what i am doing. I dont really worry about the sites below me that much but focus on making the my site better and staying one step ahead. One site in particular (site x) ranks 1st for 12 of the top 20 keywords and top 3 for the rest in an industry that is dominated by very rich companies with huge SEO budgets.

What that has to do with whether Google use clicks is beyond me. It is a non sequitor.

The reason I don't believe click based metricare used is that I don't get how is it possible to use such data. Do we really think that, for every URL and every search term, Google keeps accurate records of CTR and time on site? That isa mountain of data. Off on a tangent, is there a way to use such data that doesn't require accurate stats, e.g. can they use an approximation? How long to clicks count for? Days? weeks? Months? How many search terms do they do this for?

And lastly, doesn't clicks simply re-enforce the status quo, e.g. old sites have clicks and will always rank better because of it? and isn't it also self-referential, e.g. sites ranked one get the most clicks, therefore they continue to rank first ad nauseam?

There is always a battle with algorithms between speed, accuracy and usefulness. Soime data is great but slow, others great but useless. Click through rates, for Organic results, fail all three tests IMHO, as they would slow things down, would be hideously innacurate and likely not add much to other algorithm factors, but I am willing to hear any arguments on exactly how it could help improve results.

[gangwisch]

How can it be fraud? Whose paying/losing?
What that has to do with whether Google use clicks is beyond me. It is a non sequitor.

The reason I don't believe click based metricare used is that I don't get how is it possible to use such data. Do we really think that, for every URL and every search term, Google keeps accurate records of CTR and time on site? That isa mountain of data. Off on a tangent, is there a way to use such data that doesn't require accurate stats, e.g. can they use an approximation? How long to clicks count for? Days? weeks? Months? How many search terms do they do this for?

And lastly, doesn't clicks simply re-enforce the status quo, e.g. old sites have clicks and will always rank better because of it? and isn't it also self-referential, e.g. sites ranked one get the most clicks, therefore they continue to rank first ad nauseam?

There is always a battle with algorithms between speed, accuracy and usefulness. Soime data is great but slow, others great but useless. Click through rates, for Organic results, fail all three tests IMHO, as they would slow things down, would be hideously innacurate and likely not add much to other algorithm factors, but I am willing to hear any arguments on exactly how it could help improve results.

This does not slow anything on the serps down. the way they do the link is this way:

CODE

<a href="mylink" onclick="google('10000')"

the onclick message uses a technology called gping (you can search for it). What it mainly does is pings a server with specific data sent in the ping. However projectphp, you do have a good point on what metric would it be used for. Because it is really not an on page factor or an on page factor, it's an on-google factor. I think I may of just had a change in heart

[Alan Perkins]

Most Google SERPs do not contain Google tracking links. Also, most sites do not have Google Analytics installed. Therefore, it is highly unlikely that repeated clicking on a SERP listing will result in that listing being promoted - simply because Google does not even know that the listing is being clicked.

In general, Google is more likely to look at other factors, to do with the quality of the SERP as a whole:

1) The percentage of searchers who go to page 2 and beyond of the SERP
2) The percentage of searchers who search for something different immediately afterwards
3) The percentage of searchers who click the "Dissatisfied? Help us improve" link at the bottom of the SERP

If these percentages are too high, particularly on high volume searches, I would guess Google then looks to tweak things to put different sites in the higher SERPs.

[wassim]

Hello guys/girls,

I'm glad my question aimed to this argument about the subject. I knew that there was never a clear answer about this, it's just a question I always thought of anyway!

However, since the date I posted the question here, I started doing this trick. At that date, I started searching for a term that I optimized my site for, and I clicked on Google's result, and stayed in the site for quite a while, just in case Google don't think it's a bot clicking on the results and leaving.

I used to do that once every two days, for almost 20 days!

Now the real strange thing is (and before I canceled my domain because i bought a new one).... My site ranked #1 !!!

And the more strange thing is, I didn't touch the site in term of SEO techniques. It was ranked #2 for few months, but after I did that, the site moved up to be ranked in position #1 !!! Note that this happened after almost one week of the 20 days mentioned above.

I don't say that the act of clicking on the result increased the website position! But how do you guys explain this as I didn't even update my site at that period? Do you think this is the answer? (YES! CLICKING ON GOOGLE'S RESULT INCREASE THE POSITION!!)

Best regards,
Wassim

[Randy]

Do you think this is the answer? (YES! CLICKING ON GOOGLE'S RESULT INCREASE THE POSITION!!)

No I don't think that's the answer.

Not only is it possible that could Google tell that it was you doing that each day (browser details, ip info, the cookie they set on your computer if you don't specifically refuse it, info they can gather view the Google Toolbar if you have it installed, etc) but you're only talking about moving one position.

It's as if you think the SERPs are a static animal. They're not. Rankings go up and rankings go down. Constantly.

For all any of us know you could have been searching on a very uncompetive phrase. And for all we know there could have been one more site that started linking to yours with or without your knowledge that made the ranking difference.

This illustrates a constant problem with trying to make the data fit a theory, rather than the other way around. When you try to make observational data fit some theory it's very, very easy to mix up cause and effect. Whereas if you have a theory and then make the effort to test the theory part of your testing procedure should be to remove as many ancillary factors from the equation as possible. If you don't, you end up with invalid data because these ancillary factors can often cause a fair amount of skew.

[wassim]

Hello Randy,

what you said make sense, and I assume it is true. Again, I didn't say I went up one in position because of clicking.. I was just assuming

Come on guys, let's stop being against each other in this topic, and stay next to each other

hehe

Best regards,
Wassim