Solving the Google Sandbox

[I, Brian]

I wanted to put this up for discussion in the general SEO community. Rather than post across all the forums, DP seems to have a strong SEO base, so I thought I'd try here, and see how the discussion turns.

Google Sandbox: solved? Special report (http://www.platinax.co.uk/news/archives/2004/12/google_sandbox.html)

[SEbasic]

Really interesting read.
Thanks for letting us know about it :)

It's an interesting theory - One certianly worth discussion.

It has been mentioned in another thread (http://forums.digitalpoint.com/showthread.php?t=5608) how one of the things that may be affecting the rankings of the site, may be the lack of high PR links.

If that were the case, it would certianly back up your argument.

[zamolxes]

It is interesting and maybe true for some ... not for one of my sites though.
I already have 2 pr8, 2 pr7, a few pr6, even more pr5, etc. links (a few thousand in total) pointing to it but the site is still sandboxed (top 10 allinanchor, variable 400-600 real rankings)
I think the sandbox may be "defeated" using a combination of factors but nobody seems to really know the recipe! And maybe there is no universal solution, maybe there isn't something that can be applied to all the websites the same way. I don't think the page rank is decisive.

Also, many speculate that if there was a sandbox filter (I am a believer), one of the reasons for its existence would be to stop page rank manipulation (buying high pr links, etc.)

[SEbasic]

How long have those links been in place?

[zamolxes]

The pr8 and pr7 links haven't been there for very long (just over a month) but they should have kicked in a bit more!

[john_loch]

Hi Brian,

As you'll notice from my post at your site, I don't agree with your theory.. I explained why in the post. In simple terms your argument boils down to consistent upward movement in the SERPS once enrolled in the network.

Unfortunately that's not always the case. Auth sites can actually suffer for the dilution. It's a little difficult to measure because the 'sandbox effect' is essentially about new sites anyway, but of course if you're doubling IBL volume, it can come back into play, even for auth sites. I continue to believe 'sandboxing' is about coming on the scene with no discernable theme (actually I'm yet to see a site whos IBL's were all strictly on theme, when deep crawled by G for the first time, actually *be* sandboxed). But hey.. that's just me :)

Cheers,

JL

[zamolxes]

Hi Brian,

As you'll notice from my post at your site, I don't agree with your theory.. I explained why in the post. In simple terms your argument boils down to consistent upward movement in the SERPS once enrolled in the network.

Unfortunately that's not always the case. Auth sites can actually suffer for the dilution. It's a little difficult to measure because the 'sandbox effect' is essentially about new sites anyway, but of course if you're doubling IBL volume, it can come back into play, even for auth sites. I continue to believe 'sandboxing' is about coming on the scene with no discernable theme (actually I'm yet to see a site whos IBL's were all strictly on theme, when deep crawled by G for the first time, actually *be* sandboxed). But hey.. that's just me :)

Cheers,

JL


The question is which links exactly does google think are "on theme"? In my oppinion they often get wrong the "related" area. Wasn't it an experiment done on this forum: a page being driven to top google results without any links from relevant on theme sites?

[Solicitors Mortgages]

hiya Brian :)

after reading your report, i am interested in your quick opinion on this...

quote::""This is basic link-building practice, but the point to be made here is that you would need to emphasise building on high PR pages, rather than going for links simply in number - if you wanted to see results avoiding sandboxing delays.""

which carries more weight in the examples below (simplified)

a)links from 10 PR5's
b)links from 10 PR5's + 100 PR0's

I ask this as i wonder whether initiallty the 100 PR0's have some kind of 'negative' effect, your report seems to scream AVOID links from low PR's...surely any link is better than none,,unless it applies a negative effect?
Yours
G*M

NB: I saw your post over on SEW, I am still getting served up adsense for G*E*M's on 20K Pages ...a phenomenon that i have never heard of for sure..if i had of known i would NEVER have chosen
G-E-M as a handle....rather something more apt. (ie Solicitors Mortgages :) )

[nohaber]

high PR links count quickly; low PR links take a long time to impact That happens by default. It makes no sense, from a SE perspective to deliberately slow down low PR links even more. There's no point in doing it. Will it improve the SERPs? Will it prevent link buying? etc.etc.
Is there any reasonable benefit for Google to do it? I don't see it.

[ian_ok]

AVOID links from low PR's...

Hmm sounnds a bit like applying for that job training job and not having enough experiance.

Not good for me...low PR now...but in time, you never know...and I share my high PR when and if I get one.

a link is a link, just double check it's a GOOD link...check robots file etc....

Ian

[john_loch]

The question is which links exactly does google think are "on theme"? In my oppinion they often get wrong the "related" area. Wasn't it an experiment done on this forum: a page being driven to top google results without any links from relevant on theme sites?

ODP is a good starting point for me.
If G uses any structure to define relationships, that's a good start. That said, there's always the difficult ones, but then I tend to defer to the SERPS.

As for experiments, they do tend to yield a lot of info. Sometimes useful, sometimes not. :)

[zamolxes]

And again I say there are pages out there ranking very well without having any links from related sites - not even 1 !!

Obviously I think it is better to have relevant sites linking to you - you might actually get some traffic from them. But I wouldn't ignore links from non relevant sites.

At the moment google doesn't seem to take the relevancy of links too seriously.

[mcdar]

I have always assumed that when I add a number of low PR links that it would take longer for them to be counted as low PR pages are spidered less frequently than high PR pages.

I have never considered the time it took Google to apply credit for new links as "Sandbox".

It has been my experience to add a number of new links (both High and low PR) pointing to a page and have that recent addition of many links have a NEGATIVE or even DEVISTATING effect on the target page. (i.e. go from position #12 to nowhere to be found in SERPS!) That is what I thought the term "Sandbox" was referring to.

As the result of an ongoing (8 mos.) experiment (http://forums.digitalpoint.com/showthread.php?p=60858#post60858) some of us here have been involved with, we have shown that unvaried anchor text (all anchor text exactly the same) nor non-themed links (none of the links are on theme with the target page) have a negative effect on serp placement.

Caryl

[zamolxes]

I have always assumed that when I add a number of low PR links that it would take longer for them to be counted as low PR pages are spidered less frequently than high PR pages.

I have never considered the time it took Google to apply credit for new links as "Sandbox".

It has been my experience to add a number of new links (both High and low PR) pointing to a page and have that recent addition of many links have a NEGATIVE or even DEVISTATING effect on the target page. (i.e. go from position #12 to nowhere to be found in SERPS!) That is what I thought the term "Sandbox" was referring to.

As the result of an ongoing (8 mos.) experiment (http://forums.digitalpoint.com/showthread.php?p=60858#post60858) some of us here have been involved with, we have shown that unvaried anchor text (all anchor text exactly the same) nor non-themed links (none of the links are on theme with the target page) have a negative effect on serp placement.

Caryl


That was an amazing thread! And well done, I hope you continue to do well!

Also your keyword and datacentre tools are very useful.

[T0PS3O]

One other variable we're forgetting here - and might well explain John Loch's experience vs. other people's - is the actual targeted phrase and its competition.

I've built a new site on a new domain and all targeted KW's were pretty much #1 within 2 months with just on-Page SEO and internal linking. No so-called sandbox whatsoever. But no competition that needs mentioning either.

[zamolxes]

One other variable we're forgetting here - and might well explain John Loch's experience vs. other people's - is the actual targeted phrase and its competition.

I've built a new site on a new domain and all targeted KW's were pretty much #1 within 2 months with just on-Page SEO and internal linking. No so-called sandbox whatsoever. But no competition that needs mentioning either.

That is true, I experience it myself. I have sites competing for keywords some very competitive, others much less. It's much harder to rank in competitive areas. I haven't seen any sites getting there overnight - even with intensive work.

[mcdar]

I also find a need to question the following assertion in you article...

The clearest different between the Digital Co-op Network and my own is one of PageRank. The Co-op network relies on people offering reasonably high PageRank pages on a sitewide basis , whilst my own link building offers clients reasonably low PageRank pages on a sitewide basis.

Now, John Scott stated that he had inside information on a new system being applied by Google - that of "grandfathering of links".

The question is, how were links "grandfathered"?

My speculative answer would now be that PageRank is one of the key engines behind grandfathering itself.

This statement implies that if you use the Co-op network somehow you will receive a higher that normal mix of "high PR links".

On what evidence do you come to this conclusion?

In our experiment (linked to in previous post) we have strictly controlled the links pointing to that page. Recently someone donated links from the Co-op network.

In the Google November 25th BL Update we jumped from 469 backlinks to 3530 backlinks to the experiement page. The vast majority of the links Google now reports are from the Co-op.

You can see by this search of the first 100 links (http://www.seochat.com/?go=1&option=com_seotools&tool=7&q=link%3Awww.mcdar.com%2Fcamping1%2Fsleeping-bags.htm&result_mode=relevance&num=100&submit=PageRank+Search) reported that the vast majority of links are "Low PR".

The "*Chatabout*" links, mostly all zero PR, are from the Co-op network. In fact all of the links that Google is reporting that are coming from the Co-op are coming from VERY low PR pages.

So, I do not know how you could assert that the links coming from the Co-op are of "Higher Quality" than one would be able to acquire elsewhere.

I will NOT dispute that the Co-op is an incredible source for links, however!

Caryl

[T0PS3O]

Yes, but Caryl, isn't it still true that Google only report up to PR4-ish links?

[zamolxes]

But the links that Caryl found (from Coop) in Google show that the links from Coop have a variety of page ranks. (in this case they include hundreds of low pr)

Therefore again that somehow disproves Brian's theory.

[mcdar]

Yes, but Caryl, isn't it still true that Google only report up to PR4-ish links?

Yes, I believe that IS indeed the case now-a-days, TOP30!

SO - How are we to know how many, if any, high PR links we are getting from the Co-op?

PLUS the fact that we do know Google IS reporting 100s of zero to 1 or 2 PR links from the Co-op network shows that the assertion in the article, that the Co-op is a high concentration of "High PR" links is NOT neccessarily the case.

I just do not believe that the solution to avoiding the "Sandbox" lies just in acquiring High vs. Low PR Links.

Caryl

[T0PS3O]

In response to Zamolxes

No because Google only publicly reports low PR pages. That's why Caryl with the help of PROG finds just low PR pages. But like Brian says, in fact there are some higher PR pages in the network because it has to be sidewide. I know of PR 5's and 6es which are regarded as high(er) PR. Google won't show them with the link command but they are definitely there.

Caryl,

Shawn will know the number of high PR pages, at least the base URL's because they determine the weight. With over a 1000 sites in the network (I believe) and up to 5 ads per page, there are probably at the very least (homepage only) a 1000 PR5 or higher links in the network.

Otherwise, someone should make a tool which extracts all Y's link: results and puts tem into a PR checker.

I do agree that there are still thousands of PR0-PR3 links so it's unlikely that someone with mediocre weight will get a dozen of high PR links. Brian's theory can indeed be partly disproven with this but Caryl's assessment of the Coop links doesn't hold true either.

[mcdar]

Googles current method of reporting backlinks has made it nearly impossible to prove the effects of backlinking without performing some kind of controlled experiment.

You certainly can find no evidence from Google mere searches anymore.

Although the addition of links from the Co-op seems to have benefited our experiment page, unfortunately, the way the Co-op is set there is no way (I can tell) of knowing exactly what pages (and thus the PR of ALL of those pages) are linking to you from there.

Caryl

______________________________________

TOPS30,
Caryl's assessment of the Coop links doesn't hold true either

I am not sure what you think I was "asserting"?

I was only trying to show that the Co-op offered obviously a mix of both High AND low PRs and the article lead me to believe that the Co-op represented more High PRs.

There is no real way of knowing the number of High PR links a page is ACTUALLY getting from the Co-op.

One can ONLY assume!

Caryl

[zamolxes]

In response to Zamolxes

No because Google only publicly reports low PR pages. That's why Caryl with the help of PROG finds just low PR pages. But like Brian says, in fact there are some higher PR pages in the network because it has to be sidewide. I know of PR 5's and 6es which are regarded as high(er) PR. Google won't show them with the link command but they are definitely there.

Caryl,

Shawn will know the number of high PR pages, at least the base URL's because they determine the weight. With over a 1000 sites in the network (I believe) and up to 5 ads per page, there are probably at the very least (homepage only) a 1000 PR5 or higher links in the network.

Otherwise, someone should make a tool which extracts all Y's link: results and puts tem into a PR checker.

I do agree that there are still thousands of PR0-PR3 links so it's unlikely that someone with mediocre weight will get a dozen of high PR links. Brian's theory can indeed be partly disproven with this but Caryl's assessment of the Coop links doesn't hold true either.


You didn't understand what I meant. I know google reports only low pr links and I'm sure there are high pr pages in the network. But the fact that you can find those low pr coop links via google disproves again Brian's sandbox-avoiding theory. (He was saying that coop is exclusively or mostly high pr)

[T0PS3O]

Then I still don't understand what you mean...

Scraping (not recommended ofcourse) or manually entering all Yahoo link: result URLs and looping them through a PR checker will get you a good snapshot of PR aiming at your page. But it isn't Google's snapshot I know. And it's dynamic, changing every hour/every request. Not knowing is horrible...

[zamolxes]

I edited the post above to make it more clear. Basically Brian was saying that coop is exclusively or mostly high pr - and that is not the case - it most likely includes pages of a whole variety of pr- high and low.

[T0PS3O]

I read it differently:

The Co-op network relies on people offering reasonably high PageRank pages on a sitewide basis

I would myself have said: The Co-op network has more high PR links to offer than his own network / standard link networks.

[zamolxes]

I read it differently:



I would myself have said: The Co-op network has more high PR links to offer than his own network / standard link networks.

Would you deduce then based on the above (as Brian did) that high pr links will help you avoid the sandbox?

I for one don't.

[T0PS3O]

I do think it is a part of the solution and probably a major part but I think there's a lot more involved. The only way for them to avoid manipulation by us SEOers is to make it almost infinitely complex. There will never be one solution, especially not one that covers all our situations.

[zamolxes]

I agree - there is no one solution - the "supreme" formula - high pr links may help but will not open the door to "search paradise" .... alone!

[minstrel]

Who knows WHAT Google is reporting as PR these days, or what they are showing as backlinks? The best guess is that they are doing their utmost NOT to let anyone see PR (Google: Hiding PR (http://www.bandofgonzos.com/phpbb/viewtopic.php?t=1801)) and I suspect the same with respect to returns from the link: query.

[zamolxes]

The fact is that if you check a bit more you can find high pr pages (pr 4 and above) even now with the google "link:" command.

I recently (a couple of days ago) saw a pr 6 in their "link:" results!

[digitalpoint]

I read it differently:

I would myself have said: The Co-op network has more high PR links to offer than his own network / standard link networks.
I would agree there. The ad network happens to have pages with PR within it, but it most definitely does not rely on it or require it.

[minstrel]

The fact is that if you check a bit more you can find high pr pages (pr 4 and above) even now with the google "link:" command.

I recently (a couple of days ago) saw a pr 6 in their "link:" results!
No, you saw a page identified as PR6 -- how? Toolbar PR? Google Directory PR? Either way, the only thing you can be sure of is that those values are probably inaccurate and out of date by now, and that they are by now probably intentionally out of date or inaccurate...

[zamolxes]

I can only say what I saw. I mean, everythingelse is speculation. And not necessarily correct.
For example: I recently (a couple of months ago) started 2 new websites, just a few days before the last toolbar pr update. They were crawled literaly just before the update and they only had by then 3-4 average low pr links pointing to each of them.
A few days later, after the update,the toolbar was showing for one pr2, for the other pr1. I would say that is pretty accurate - nothing hidden or out of date in my experience.

[minstrel]

I can only say what I saw. I mean, everything else is speculation. And not necessarily correct... nothing hidden or out of date in my experience
Well I can only say that your experience probably isn't the norm -- granted much of what one finds in forums (fora?) is rank speculation and blind guesswork, but nonetheless relying on Toolbar PR to provide you with accurate information about PageRank is the epitome of blind faith, if you ask me.

[zamolxes]

I partially agree, but I wouldn't push it to the extreme.
What I forgot to say is that I had an even newer site (still crawled before the toolbar update) whith more and better links pointing to it - I guess it just missed the update cycle somehow; it is still pr0. So, I guess if google will update the toolbar quarterly and given the growing amount of links to the site, at some point the toolbar pr becomes .... inaccurate.

What I mean is there might still be a logical explanation. And many forums cultivate some kind of paranoia because of things misunderstood or missinterpreted.

[gchaney]

I contend that in fact your analysis is flat wrong (but I do so with love and admiration!! ;) ) You are missing the most important aspect of PR in your theory.

I have seen a site which is new that ranked very well. How? One mortgage site was a part of a "Larger" network of sites. No reciprical obl that I could find with All i/b/l from "Authority" (One PR 7 for sure) and regular old range of links from a large network of sites. I didn't note the domain during my review as mortgage was not my focus at the time and it was provided by someone doing their own research focused on PR only. At that time, I felt and still feel it has nothing to do with PR and evaluated the sites offered that confirmed my own premise.

To my own methods which led me to this. I tracked for this key phrase "My Special Widget" on a regular basis, manually last 4 months up until beginning of last month, and with program after....never noted two specific sites in top 100 up until the last 60 days (Hey I had to go deep to find my site...lol) and absolutely watched them climb with every update; I Look at the b/l. When I analyzed both these sites (have since chased there links this month to...lol) after watching them jump up from nowhere last two months to #3 and #5 respectively, you might be surprised at what I found.

Site "A" had a mix of crap unrelated PR0 and a handfull of "Authority links" with "Authority" outbound links , with no "exchanges" to speak of excluding one with a authority Gov't site. This was when I did my initial analysis at the mid point of October and "A" was ranking in the 90's. Since the original Yahoo & G evaluation, tons of new i/b/l's links have shown up for site "A" with still no reciprical links much less the keyphrase in the anchor for most.

Site "B" came from nowhere as well. Same term.

ALL of the outbound links are only to "Authority" sites. The remaining are all one way i/b/l - Some will say age of site/domain as they are older 3 & 7 years respectively, however, both these sites have aquired a large number of NEW links (over not less than 800 each) a very short period of time as I've watched them grow, and as they did, so they grew in the serps.

(As a note, I evaluated the sites as one is gov't on is not and relates to my topic. I linked to both of them under a "competition" page that describes what each does....lol)

Why did they grow in the serps so fast? You have to follow me on this, and match it to PR principles in order to catch it. To me it now just seems to obvious and actually ironic...lol.

The minute G indexes a link from your site to another site, you have cast a Vote for that site. Every one has noted the significant delay in PR updates since I believe about June. I think this all plays in together...... Follow me now...and remember the PR principle.

So when I cast a Vote, that vote says site "B" is important. If G then follows that link at a later time only to find the site you're linking to has a link back to you, it sets you up for the sand box for those links. How?

I don't believe that G is hiding PR intentionally. I believe G may actually be doing detailed link analysis looking for reciprical links prior to doing a "Full" PR updating. This would certainly take time and require a full recalculation based upon comparative data which would explain the movement from monthly to quarterly. They must filter out the results.

Why would G bother filtering results???? Think interms of G doing a "Vote" recount in a disputed election..... and keep following

During this analysis, G must discount these new reciprical links to no value or to a percentage of value...similar to a hanging chad. I believe this is relevant to PR. I think the higher the PR, the less value the vote has. Why? Think in terms of SEO principals. A link to a high PR site is a principle part of SEO. However, a recipricle link to a PR0 site is considered a waste of time and generally not done unless desperate for a link, or the PR0 site actually has some pretty cool stuff on it. Think like an engineer....thinking like an SEO The Engineer know you want PR, so why would the new algorithum give you credit for that.....lol.....keep following...you'll get how this all works out....

So, what do I base this premise on? The PR principle itself and how each vote is counted. What do I mean by this?

In principal and fact, yes PR authority on a one way i/b/l definitely gives authority to the link, I also contend stongly any PR i/b/l can help (including 0's) as well so long as they are not reciprical and are a "True Vote" cast for your site. I believe related sites create stronger links, but are not necessary to count as a "Vote" for you.

This would 100% support G's premise of Site "A" pointing to Site "B" gives site "B" a "Vote" of importance. There is no way G can maintain the credibility of PR and the ranking algorithum any other way.

So, when you're out exchange links, you are defeating the most important aspect of G's PR algorithum. It's not the PR itself, but the "Vote" and how it is cast that makes PR important.

A standard technique of SEO is "creating" a vote. This goes againt the "Democratic" principles of natural PR and a sites natural "Votes". I truly believe the engineers at G hate SEO's for manipulating them and the precious PR. So saying, they are always looking for ways to defeat our tactics while not saying what they are doing.

This new game they have chosen to play is nothing more than a "link exchange" penalty or what I consider a "Voter Fraud" jail term.

This is what I based my premise on with my own theory a couple weeks ago in another forum that got me ragged as well...lol. That the minute you start exchanging links, your new links are "Jailed for Vote Fraud" (though I did not use this specific term). It is why I joined this co-op and believe it will make a difference.

I also believe this is why older sites are not affected as they are typically not link exchanging anymore or have any need to cast "Fraudulent Votes".

I don't believe there is anything more to the sandbox than that. I think we're all just over analyze it and forget the premise of PR to begin with, which many of you know by heart.

So, if you're in the co-op, and you are out "Exchanging Links" or committing "Voter Fraud" the benefit of the co-op may actually not be apparent in G.

So, no matter what anyone says....lol.....if you look at what PR is suppose to represent, and then look at what a Link Exchange is in relation to PR, well...you have now identified the reason for your sites "Jail Term". It's nothing more than a penalty for committing Voter Fraud....and during an election year to! Isn't that just ironic....lmbo

Cheers

[minstrel]

IF (and that's a big IF) Google is penalizing or discounting "reciprocal links", I don't believe it is as simple as linking to a site that links back to you. It MAY be a funtion of relevance of that site to your site... but not based on mere reciprocity. That would make no sense at all to the PR principle.

[zamolxes]

Sounds interesting and again may be partially true in the sense that google somehow might give less weight to some reciprocal links.
However they still show many reciprocal links using the "link:" search - in fact it doesn't seem to me like they filter them out any more than other links.

On the other hand there can be good legitimate reasons for sites to link to each other (other than pr and link popularity) and I'm sure google knows that.

[T0PS3O]

So does your Vote Fraud apply per page or per site?

I mean if link X on page 1 on site A links to page 2 on Site B but site B has link Z linking to site A's page 90, is that fraud?

[gchaney]

G does not show all the links and I believe that is again intentional. Google does not want it easy for SEO's to identify links for SEO purposes. Remember, links are suppose to be natural VOTE's a natural vote does not require you to know where all the votes are coming from....lol

BTW, I didn't say they removed the links, I think they take away the value.

T0PS3O So does your Vote Fraud apply per page or per site?

As this is a theory based upon the intent of PR vs what we are seeing, I believe yes, G recognizes there is "value" in "A" site linking to "B" site. My rambling was long enough...however, I do believe there is a "Factor" as a percentage of votes vs recip links vs anchor text and consequence.

The adjustment most definitely must be complicated. This was simply a general analysis of the root cause and affect as many new sites start out doing mass link exchanges forgetting one way linking.

[I, Brian]

Crikey – this thread grew fast. :)

The point about the low PR links was *never* that low PR links can harm a site – merely that they will not be counted for a delayed amount of time. The expectation would be that they would normally be counted and their “vote” applied without a forced delay period.

While I very much agree with mcdar that there is the expectation that low PR links will take longer to index – the problem is that even after these links are indexed, they are not impacting the SERPs when they should normally be expected to – my understanding is that this is entirely where and why the Google Sandbox has become an issue of such discussion.

I have to confess to not having read followed the previous discussion on off-topic linking – but my commercial experience is that off-topic links have value and in number can be extremely effective. Obviously, though, some degree of on-topic linking is preferred in the mix.

Also, the point that adding links fast being “devastating” for rankings might well be addressed by issues of far too similar anchor text, which is a going concern in SEO. Part of the general SEO procedure often involves “naturalising” the links, which also allows for a wider scope of search terms to be impacted.


So, I do not know how you could assert that the links coming from the Co-op are of "Higher Quality" than one would be able to acquire elsewhere.

The point there is that the co-op network includes a number of higher PR pages. Certainly there are a lot of low PR pages involved – we’re often talking about sitewide linkage, so of course they will be in number. However, the co-op network also involves a lot of higher PR pages – index pages, main navigation pages.

According to the suggestion in the report, it is precisely these “higher quality” “recommendations” that Google is taking positively into consideration, and my contention is that this widespread “recommendation” by so many higher PR pages across a number of sites is precisely why the Co-op network is allowing links to impact the SERPs with almost immediate effect.

After all, the big question is why the Co-op network avoids sandboxing – I’m simply putting forward the PageRank aspect as a contending possibility.

Certainly John Loch’s comments ring true, too – the Co-op network is no substitute for commercial SEO in highly competitive areas. But, again, I can set up a few thousand low PR links, see them indexed 3 months ago, and not see the anchor text impact even a non-competitive term. But the moment I add that to the Co-op network, my keyphrase was top of the SERPs within a week. That’s the big incongruity I’m trying to address.


the fact that we do know Google IS reporting 100s of zero to 1 or 2 PR links

Part of the problem is that toolbar PR is not being accurately reported – there is definitely a good lag on the actual value of PR being given, which was mentioned in the report.

Also remember that dynamic forum pages traditionally never show PageRank on the toolbar – that has always been an issue. Even when the backlinks were only supposed to show PR4+ backlinks, dynamic forum threads with .php extensions have always come up as zero. It was only their showing on backlink updates that suggested they had any PR at all – but even then, if PR updates constantly, as is suspected, then those threads in backlinks will certainly be showing an out-of-date PageRank anyway.

As for Google’s attitude towards reciprocal links – certainly I’ve seen *a lot* of debate centered on whether Google is even counting such links (the old “links.htm” controversy comes to mind). However, the majority of what I’m referring to is one way linkage. I actually mentioned recips as something of an addendum, but I hope that discussion point does not distract too much from the main contentions I’ve raised.

[longcall911]

Hi Brian,

Taking your thought a step farther, isn’t it true that high PR sites are crawled more frequently and more deeply than low PR sites? If so, it stands to reason that their pages would also be analyzed (links, keywords, titles, etc) more quickly than low PR sites.

That might explain the coop’s faster impact on SERPs.

/*tom*/

[john_loch]

Hi Brian,

Taking your thought a step farther, isn’t it true that high PR sites are crawled more frequently and more deeply than low PR sites? If so, it stands to reason that their pages would also be analyzed (links, keywords, titles, etc) more quickly than low PR sites.

That might explain the coop’s faster impact on SERPs.

/*tom*/

Actually longcall, that's where things get interesting. Very true that high PR pages cop a crawl daily if not hourly, but I just saw a link newly introduced to the network appear across 30+ sites (well, 30 that caught my attention out of 5000+), all of which have PR0. It was introduced i think 3 days ago.

The interesting apsect of this is not so much the PR for me. For me it's given rise to an alternate possibility. When someone lands at a site, sooner or later the internal linkage lands them at a hub, usually the home page. What if a link from the internal hub (which will obviously by way of link structure also have the greater PR) actually yields greater benefit to the recipient.

By default, you could have PR0, but of course, be breaking the latest research news or whatever. Gfodder big-time. Is it not possible that this internal hub is primary, with the PR a natural second ?

Above and beyond everything else though, Occys shaver wins out. At the end of the day if the process of elimination leaves you with only 1 solution, no matter how unlikely.. ;)

I still believe theme is very significant, because just as you deal with local hubs, thematic hubs (hub/auth sites) also make the most sense. I suppose for me it's always a matter of remembering that the technology is designed to assess human information structures, and not VV (although it certainly keeps me busy) :).

Cheers

JL

[longcall911]

Gfodder big-time. Is it not possible that this internal hub is primary, with the PR a natural second?Yes it is very possible.

I think I’m following your hub example, but am at a bit of a disadvantage because I don’t know the coop at all. My basic understanding will probably suffice though. My question would be, what physical, measurable characteristics are typical of a coop hub?

Is it possible that they demonstrate large blocks of content that change frequently, perhaps hourly? Now if those blocks were structured as we would expect news to be, that is with a heading here or there, some content related to the heading, and maybe a scattered date or two, we have a number of attributes that are very measurable.

We all know GG prides itself on being able to get news content indexed quickly. I think it is safe to assume that GG follows a set of rules to identify a *news* site as opposed to humans doing it on a site-by-site basis.

So, does the coop happen to pass most of the heuristic rules therefore giving the appearance of a news site and therefore receiving the benefits? That would explain rapid indexing of PR0 pages, and maybe even the indexing of brand new pages. Perhaps news sites and portals are exempt (to some degree) from sandbox periods. That gets right back to Brian’s original sandbox point.

It would be pretty funny if GG thinks the coop is a monster news portal. :-)

[digitalpoint]

Personally, I don't think there is any sort of sandbox for pages. Just sites/domains as a whole. That's why topics in this forum rank high the day after they are posted, and new random junk on my personal blog (and everyone else's I would assume) always ranks high.

[chachi]

Well, I think that most of what Brian had to say was definitely food for thought. I have always felt that there is some kind of sandboxing or incubation period for the links pointing to a domain or page. After reading through this thread and sniffing the paint fumes from the paint outside, I am thinking that maybe the filter G is using (if they are) only allows a site to receive a certain number of links (votes) based on the time in the index. We all know that new sites will not rank for semi-competitive to competitive terms right away...but, we don't know why. Perhaps your new site is only allowed to get the voting power from say 100 links per month in the G index (for example)? Perhaps the number of pages with x amount of content on them bumps that number up....as it seems large (5k+ pages) new sites can build links (with the link: operator) more quickly than say a 20 page site?

If something like that were true it may explain why DP.com is able to rank so quickly for new terms. And, why DP.com is able to really benefit from the coop links (as it presently shows 90k+ pages (non-api)) and a gazillion links.

Incubating the links and sites that way would make sense to me if I were G. I would want to encourage new sites to be included in the index, but not want them to be able to be the BMOC for a while. I would want to feel out the sites first and see how the behave.

Hopefully this makes some kind of sense...I need some fresh air. :)

[digitalpoint]

To be honest, I don't really think the coop ad network bypasses the sandbox. There are many sites in the coop ad network that are #1 for allinanchor but no where in the normal query results. For the ones I happen to know about, the only thing they share is that they are all domains that were registered within the last few months.

[I, Brian]

What if a link from the internal hub (which will obviously by way of link structure also have the greater PR) actually yields greater benefit to the recipient.

The question of hub has certainly been an idea - but if we go all the way back to the original Google paper, "PageRank" and "authority" were originally supposed to be linked conceptually anyway.


Personally, I don't think there is any sort of sandbox for pages. Just sites/domains as a whole.


Indeed, Barry disagrees too:
http://www.seroundtable.com/archives/001218.html

So I've set up a quick little test for a page on a very new site, using the Co-op network: (third post down)
http://www.platinax.co.uk/community/thread1408.html


To be honest, I don't really think the coop ad network bypasses the sandbox. There are many sites in the coop ad network that are #1 for allinanchor but no where in the normal query results.


Certainly there is going to be lots of low PR linkage involved, which my initial idea suggests isn't going to impact SERPs anytime soon.

However, if PageRank is a major factor, then the larger number of higher PageRank pages are enough to help move sites up to some degree in the listings. Obviously, the degree of active competition in the SERPs is likely going to be a factor.

As before, the key question I'm trying to address, though, is why links on the Co-op network can have any kind of relatively quick impact on rankings at all - certainly when other links may not.

[digitalpoint]

It may just be as simple as pages with PageRank are spidered more regularly, so they are seen faster. Check allinanchor for eharmony for example. #1 and #2 (above even eharmony.com). link: function shows a ton, but the domain itself it relatively new. Another similar example is hummer dealer (link: doesn't show much on that one because it's very new to the coop). But I expect they will be top 5 on their terms at the very least whenever the sandbox is done.

They also have some good solid static PR6 links as well (not from the coop).

[Elee]

Can anyone tell me how long a site is typically "sandboxed"??

[I, Brian]

Personally, I don't think there is any sort of sandbox for pages. Just sites/domains as a whole..
Actually, if Google are using an automated WHOIS query to give weight to site links based on length of time domain registered, then that might also explain the exact same thing.

So the issue of PageRank could be entirely incidental - in other words, older more established sites will generally have respectable PR anyway.

[darksat]

Actually, if Google are using an automated WHOIS query to give weight to site links based on length of time domain registered, then that might also explain the exact same thing.



They wouldnt need to, they could just take the age of the site from when googlebot first found it, it mightnt be as accurate in the technical sense but it would be as effective.

[longcall911]

. . .in other words, older more established sites will generally have respectable PR anyway.
I would be inclined to credit site age more so than PR, although I’m not suggesting that PR is unimportant. I think it is all about *trust*. That is, Google’s development of a level of trust, over time.

If I were GG, I would trust no one. I’d assume that every new site was out to deceive me. So, I’d sandbox every new site until I knew more about it. If I started to see links from other sites, and saw regularly added content, and if I saw general stability, I’d let it out of the box and watch it closely.

Over time, if all seemed normal, I would trust it more. As the site matured, and once it gained my full trust, I would take clear notice of the sites (pages) it links to.

If another new site came along, and if the first (fully trusted) site linked to it, this would boost my trust in the new site. In other words, the new site would *inherit* my trust. I would therefore crawl this new site more often, and I would analyze its pages faster, and I would remove some of the ‘rank inhibitors’ I had placed on it, allowing it to rise to its natural position in listings.

I think that if Google trusts a site, it can do a lot of things (like SEO) and not be penalized. I have a close friend who I trust completely. I might say to him, “I’m going away for the weekend. Here are the keys to my house”. And because I trust him, I trust his friends.

His friends inherit my trust. This is the exact behavior I see in Google, although I’m still waiting for keys to the house. :-) The point is, the co-op may be seen as a group of highly trusted sites.

[I, Brian]

A very fair point indeed - and the time the site has been in Google's index, as opposed to a WHOIS query, is definitely an angle I overlooked.

[zez]

I have seen many Sandbox theories, but the idea that Google start to 'trust' a site after certain amount of time in my opinion is most propable. I have a site, which after 1 year suddenly has started to have a very good SERP on many competitive keywords. The site is only PR3 and does not have many links. Within this site I have page which was recently added and it has PR 0 but this page is doing is suprisingly very well.

[4Comparison]

While reading the entire thread, age keeps coming up as an under-rated variable.
When developing a new site (non-commercial), most links are acquired from others eager to exchange links, and are mostly new sites leading to a delay in the effectiveness of the links.
Could the Co-op's substantial growth over the last 6 months or so be the influx of older sites, thus giving a boost of "trust" to the network?

Additionally, on another thread, someone wrote of their disgust with the sandbox and moved their entire site to a sub-domain of an old, established site and got almost immediate high rankings.

[Catfish]

1) I agree with Shawn that the sandbox is URL/Domain related. We run a large network of sites that are all hosted on unique class C ips and have excellent page rank. We interlink these sites using triangular exchanges and the resulting page rank is usually a 5 or 6 to the homepage of any new site we put up. Typically our sandbox is running about 3 months right now for most new sites.

2) I don't believe that reciprocal links are being penalized or most of our sites would have dropped dramatically in the SERPs. People's speculation of how things should be or what makes sense for a search engine to do, often times clouds their view of what is happening in the real world. No matter how many times Doug (from ihelpyouservices.com) and some other white hats try to tell me that reciprocal links are a waste of time, the proof is in the listings.

Most people think page rank is dead too. But its not. I betcha Shawn doesn't think page rank is dead (hi Shawn). I belive that page rank is why Google bombing can still take place in the example where someone links to Google or some other high PR site with a link like www.google.com?keyword and suddenly Google ranks highly for the keyword.

But regardless of the page rank argument (sorry I was drifting), reciprocal links are not dead and they still work. The trick is to know how to identify the pages that will help the most and try to acquire those links. I have seen lots of forum posts where people are submitting to every directory under the sun thinking that will help their backlinks when most of these directories don't pass PR because of their linking structure. If TSPR is in play, it would seem to me that the best link partners are ones listed in DMOZ in a related category to your own. Anyway, thats been my experience. Hope it helps.

[longcall911]

Could the Co-op's substantial growth over the last 6 months or so be the influx of older sites, thus giving a boost of "trust" to the network?If you choose to consider the concept of trust, then one could further clarify the issue by saying that age is in fact important, but that a site’s behavioral pattern over time, and the trustworthiness of its link partners also figure in.

Interlinking between older sites that seem to have honored Google’s golden rule “don’t do anything that attempts to deceive us in any way” (the co-op may fit this profile) may receive the benefit of Google’s trust. . . higher rankings and shorter sandbox periods for newly added sites/pages. (trustFactor = 1.0 as opposed to .85 for example)

[webvivre]

Does this mean......

a) Non relevant reciprocal Links from older sites / authority sites are acceptable in Google's eyes?
b) We should be looking for reciprocal links from older sites because they effectively carry more weight?

Does old site = good site in Google's eyes? Don't think so.....

[longcall911]

Does this mean......
a) Non relevant reciprocal Links from older sites / authority sites are acceptable in Google's eyes?
b) We should be looking for reciprocal links from older sites because they effectively carry more weight?
Does old site = good site in Google's eyes? Don't think so.....
First, please understand that the issue of ‘trust’ is pure speculation on my part. But, after paying very close attention to Google’s behavior over the last 8 months, trust has surfaced in my mind as a very possible key.

In response to your:

a) I would say yes if these older sites are trusted. I think that links from trusted sites would increase one’s trust factor (assuming there is such a thing as a trust factor). That alone, would probably have some impact on SERPs and amount of time in the sandbox. I believe that relevant links from more than one trusted site would have significant impact.

b) Older sites would be a starting point. I think you want trusted sites. Since trust takes time to build, I think it is more like that an older site would be trusted than a newer site. I think an old site with hidden text or keyword stuffed alt tags for example, would be near worthless. It might even be a PR6 because of its shear number of links, but still worthless because it is not trusted. I’d look for ‘good citizen’ sites, clean of any SEO ‘tricks’ and with backlinks from other clean sites. Those sites probably rank well for main and secondary keywords.

No I don’t think old site = good site. I think the main factors that determine rank are relevancy to the search terms, trustworthiness over time, and to some extent, link popularity (PR). In other words, the formula might be: relevancy x trustfactor x PR = site_weight for a specific keyword/phrase. If so, the next question is how heavily is each of the components weighted.

Lowering the importance of PR in that formula could have a significant impact on SERPs. Increasing the importance of trust could have the same impact. I think we are already seeing these tweaks.

[kokopoko]

The suggestions to avoid the sandbox are pretty much what we already know: get high PR links back to the main pages of your site. It will help avoid sandboxing and if the site gets boxed anyway, the links will help get it back up and running faster.

[longcall911]

get high PR links back to the main pages of your site.What I am suggesting is that high PR may not be the real qualifier.

A page can reach PR7 with merely enough links. What if some number of those links are to sites that have been penalized either now or in the past? That would not necessarily effect PR at all.

But if there is a trust factor, and if that factor were lower than average for this PR7 site, a link from it would do you little or no good. Worse, I contend it could actually hurt by lowering your own site’s trust level.

[lilquit]

Can anyone tell me how long a site is typically "sandboxed"??

Yeah about how long again?
thank you

[CanadianEh]

Well written article Brian. Started some good discussion. Your "high PR: link is definitely a valid theory that deserves further investigation.

Here are my 2 cents (note: I am not an expert and 2 cents Canadian is worth less in US and Britain) :)

Google has noticed the tendency of webmasters for creating new sites in order to rank well. This happens every minute of the day. Not wanting to encourage this behaviour, it would make sense to remove any form of instant gratification. Think of it, if you new that you could create a site, add 500 links to it and have it go top 5 in a couple of weeks, you would do it more often.
Google prides itself on adding new content quickly. Most people don't notice if a new site is not listed high, unless it is newsworthy, and here is a business reason for Brian's logic. For example, if a million people did a search looking for the "site" about "Janet Jackson's wardrobe malfunction" and didn't find it, they might curse Google and look elsewhere (not a real example). Google would want to prevent that so they might want to give exclusion to sites with links from high PR sites like USA Today and the New York Times.


I noticed a couple people giving examples of how Brian's theory didn't apply in their situation. Please remember that it also makes sense for Google to assume spamming if the link text coming in is too similar. Could be the reason for the exception.

[lilquit]

How long does the average sanbox last again?

[d360]

I'm coming in late to the thread but around April I had about four new domains. One domain we completed a 30 page site the others were one page placeholders. All using the style of coding and SEO practices the 30 page site actually took over a month longer to get PR than the one page sites. They were all linked from the same PR 7 site. They all have PR 5 and 6 now but it took the first couple domains 3 months to show up.

I've done many tests and if we use our PR7 site we can get new pages on the same domain a PR within about 2weeks vs. the new domain taking about 3-4 months

[d360]

How long does the average sanbox last again?


My guess is 3-4 months based on experience. How many threads does it take to get links below your messages in digital point forum?

[minstrel]

How long does the average sandbox last again?
If there are cats in the neighborhood, about a day and a half, tops.