No Follow to Homepage - Why are sites doing this?

[irldonalb]

I've recently followed matt cutts' instructions to highlight no-follow in Firefox [mattcutts.com].

Now when I browser the web, I see what other sites are doing.

I'm baffled by what I believe is misuse of the tag.

- A leading SEO blog / magazine has the tag on all links their homepage. Also within a page, they've the tag on some links to category pages. This ignores the "first link counts" logic spoken about previously here and at pubcon
- On a leading tech magazine (PR8), they've a no follow tag to their homepage from every page.
- Top UK portal / newspaper (PR7) has no follow tags to all their categories including homepage

Do they know something I don’t know?

Thanks
Donal

[tedster]

I doubt it very much. I think sites are trying to be clever but making problems for themselves instead. This kind of nofollow abuse may be why Matt posted the recent video we're talking about here: [webmasterworld.com...]

[steveb]

It sounds like they are nofollowing bad anchor text duplicates, which is fine and sensible.

[TheMadScientist]

fine and sensible

Until they realize nofollow on one link from a page to a page nofollows them all on that page to the other page, so it does not help to nofollow 'duplicate' or 'undesirable' anchor text hoping the other anchor text will count or be credited as the anchor text, because nofollow on one link to a URL on a page = nofollow all links on the page to the nofollowed URL. (Or so I've heard anyway...)

[steveb]

That's not true.

[irldonalb]

Steveb - I think TheMadScientist is correct.

It was one of the big things to come out of pubcon last yr.

Plus it was asked to Matt Cutts. he didn't deny or confirm it. But he did say it doesn't work as expected.

[tedster]

I think we're getting tripped up on the wording here. As I understand it, this is what happens:

1. When there is a nofollow link on a page, the other, dofollow links on the same page still "count", but they DON'T get any part of the PR back that the nofollow link would have had. Originally they did, but not anymore.

2. At PubCon, Bruce Clay discussed some testing he did. He tested two links on the same page that point to the same target URL. Bruce said he discovered that only the first anchor text counts.

So yes, these sites described in the opening post may be trying to get that first link ignored hoping that a second link somewhere on the page gets its anchor text used as a ranking boost.

[TheMadScientist]

Hmmmm...

I was going with tedster's post in the following thread (the 2nd to last post), but am testing now: [webmasterworld.com...]

[tedster]

Thank you, oh mad one! I haven't actually tested the #3 point that Bruce made -- and truth be told, I forgot about it because I don't use internal nofollow links for this kind of thing:

3. If you use nofollow on even one of the multiple links, they all stop passing PR.

From a Google point of view that makes sense. If one occurence of a linked URL is "not trusted" why would a second one be trusted, anyway?

----

This whole area has been one of conflicting test reports from various people. For example, some say they've tested using a page fragment identifier on the second link and that it does pass anchor text influence in that case. Others have reported that Bruce's test doesn't agree with what they see in their tests -- but still several say their results agree.

Whatever is the case in any moment, it could also change in any moment. So if you are working with this kind of approach, I'd say you need to test it frequently. It is the kind of tactic that Google usually doesn't want to see working.

[edited by: tedster at 6:42 pm (utc) on Jul 3, 2010]

[TheMadScientist]

If one occurence of a linked URL is "not trusted" why would a second one be trusted, anyway?

That's exactly why I didn't bother to test it either.
You posted it, Bruce said he tested it.
It makes quite a bit of sense.

I believe you probably cited him correctly and he obviously tested it, plus it makes absolute sense to me, and I really don't care, so why even bother with checking? Now I've got the curiosity issue going... Since it's a 'Google Thing' I wouldn't be surprised if there are times when it does and times when it doesn't, because about everything they do is relative to other factors anyway.

[Robert Charlton]

No Follow to Homepage - Why are sites doing this?

I'm thinking it's because the sites heard about Bruce's test that the first anchor text counts, and that they don't want that anchor text to be the word "Home" in their breadcrumbs or whatever their logo-link-to-home alt text says.

They might even have a keyword anchor to Home somewhere in the paragraph text of the page, since they've also heard that global nav text (in the page template) may count less than a contextual link does.

It's also likely that these same sites haven't heard about #3, though... Bruce's test that if you use nofollow on even one of the multiple links, they all stop passing PR.

I'm curious, btw, how Google treats these various ordering factors when you use source ordered content.

irldonalb - Since you've spotted these nofollows, I'd be interested in hearing whether there is a targeted anchor to home used somewhere on these pages instead.

[steveb]

"If one occurence of a linked URL is "not trusted" why would a second one be trusted, anyway?"

Because such a practice would be idiotic. Such illogic means you could ruin the PR distribution of any blog or forum site you wanted simply by adding the site's main page link (or their entire navigation) to the blog comments on any page.

"It makes quite a bit of sense."

It make no sense at all. It would be pretty hard for Google to do anything more bizarrely destructive.

[TheMadScientist]

Absolutely genius point!

Except for the fact using your theory would only remove the PR flow from a specific page, not the entire site or every page on the site, so to really 'ruin the PR distribution' you would have to do it on the entire website, or a significant portion at least. Seems like a 'small trade off' for Google to prevent PR sculpting, IMO... And if it is applied situationally as I suggested, then it might not even do that.

[tedster]

you could ruin the PR distribution of any blog or forum site you wanted simply by adding the site's main page link (or their entire navigation) to the blog comments on any page

That certainly is an edge case, steveb. If Google is doing this for nofollowed blog comments, it would disrupt internal circulation from that one page. But to do any significant degree of damage you'd need to add such a link to the comments on many, many pages, right?

The tests that Bruce Clay and others were doing involved multiple links in the content area. Google does segment the page in their algorithm, so I assume that blog comment areas are treated differently, no matter what.

[steveb]

Nofollow only makes sense granularly, otherwise the illogic applied above ("If one occurence of a linked URL is 'not trusted'...") would make just as much sense sitewide, because it would be the height of stupidity for Google to start thinking like: "you don't trust it once on this page, so we will force you not to trust it elsewhere on this page, but we will allow you to trust it on other pages on your site."

Next up for the FUD spreaders, an explanation of why Google still show such links as backlinks.

Google may do all sorts of crazy stuff, including experiments, but crazy will never "make sense".

[TheMadScientist]

You're right steveb: Bruce Clay, tedster and myself are stupid...
Feel free to troll along, nothing more for you to see here.

[tedster]

I don't feel like steveb is trolling. I appreciate his differing point of view, which he clearly feels is correct, and he's making me examine some things more closely. If I change the way I see this issue, I'll post about it.

[TheMadScientist]

Fair enough... But the tone of the posts he's making IMO are about derogatory. I love differing opinions and discussions, but the overly critical, absolute manner when testing has been claimed to show other than what he is saying must be the case just gets old, and I think it was one of the biggest gripes in the FeedBack Forum. IMO his posts definitely don't encourage others to join our discussion in the least... I know if I wasn't a regular there's no way I jump in to share my thoughts in a thread where someone is posting as the absolute authority on the subject and implying those who believe the testing cited are spreading FUD, regardless of the fact results the testing others have cited shows the opposite of the position he has taken.

[steveb]

There is one thing for sure... it makes no sense at all to handle nofollow that way. Zero. It's just illogically inconsistent.

On the other hand, it is a whole different thing whether Google does it or not. Google is capable of really dumb things, but there is zero evidence that they do this this dumb thing.

In any case, this is just about the easiest thing to test. Build a page, link to it twice from ten or 100 or 10,000 pages with two links, one of which is nofollowed. See what happens two days from now.

What happens today is the page is indexed, the backlink can be diplayed by Google's link: command, and the page ranks. And all of that is logical and consistent with everything Google has said about nofollow.

And TheMadScientist, please get a grip. I said it would be stupid for GOOGLE to do this. You believing and stating unequivicolly something so illogical is not what I said is stupid.

[TheMadScientist]

Thanks for the post to actually clear things up a bit more about your opinion rather than just saying it's stupid to think they could do it or it's been done by them or the previous testing is not even possible, because you're right, sometimes they do what I think are really dumb things, and I also think Bruce Clay probably tested it before he announced it at PubCon (I've talked with him a bit before, and happen to know he really does test when he says he does)... IOW: I'm fairly certain Bruce didn't simply make it up or pull it out of his a**, because that's not how he operates, unless he's changed his MO.

Anyway yes, my testing is already in place, and has been since minutes after I posted I would start testing myself even though it does not really effect me or the sites I work on.

steveb: Sorry if I misunderstood you... It's happened to me before and I think one of the biggest issues to 'tone' or understanding around here is sometimes we type things we could say and they would be understood by tone and inflection, but with type we don't get that, so sometimes it's difficult to 'get' what someone is saying or why without more information. I know sometimes I come across as more of an ass than I really am to people, so, again, my apologies if I misunderstood, or read-between-the-lines too much.

[edited by: TheMadScientist at 12:46 am (utc) on Jul 4, 2010]

[steveb]

I suppose should also have added that building two more pages would be proper testing... one with two nofollow links to it, and then the third with two normal links to it. And link to all three with the exact same number of links in the same area of the pages.

[TheMadScientist]

Yeah, to really be accurate it should probably be tested with a blog and a content site, internal and external links, and there should probably be a minimum of 4 test links on the pages used for testing. One in the bread crumbs, one in the main nav, one in the text (content area) and one in the footer with the nofollow rotated through them, and on the blog another link in the comment section should probably be added... This is one of the reasons I didn't bother before, and I think they could certainly be treated differently between internal and external. It makes the most sense (in a nonsensical way or course) to apply nofollow one = nofollow all on the external links on a page, but not to the internal links on a page.

So I could fairly easily see the following being applied, whether it makes complete sense or not:
nofollow on one external link on a page = don't credit any links to the external page
nofollow on an internal link = pass the PR anyway as long as there is one link

OR

Even reversing it on internal links, and disregard the nofollow all together and if there is one follow link to the page, treat all the links as live and score / credit the text the same way, because the only reason I could see to nofollow one internal link and not another is sculpting, unless your blog comment idea is used by someone, and then it's still 'the right thing' to go ahead and count them all... Why else (besides sculpting link text or PR) would someone nofollow one internal link and not all of them? An accident? Count 'em all sounds right... On purpose? You're trying to game, count 'em all... Someone else did it? Count 'em all...

Of course, if you wanted to apply an indirect filter or penalty to the sites trying to sculpt with nofollow I couldn't think of a much easier way than to let them decimate their PR flow by applying nofollow one = nofollow all even on the internal links on the page...

But, there's quite a bit more testing than I care to do to find the answer to a question I don't really care what the answer is to, so I'm not going to do the extensive testing I would if I really cared, but it makes some sense they would treat internal and external differently, but IMO it's really not worth the risk of having them change their mind and if for some reason I nofollow an internal link on a page I nofollow them all, because then it removes this question from the equation.

There's better ways IMO to get the text I want counted in the right place or keep a page from receiving PR from internal pages than leaving it to a 'Gwhim' that could change at any moment.

[edited by: TheMadScientist at 2:13 am (utc) on Jul 4, 2010]

[TheMadScientist]

I think my answer to the original post is:

They're either trying to 'show how great they are and how much they know' or they're trying to sculpt text counted or links, and the bottom line IMO is... Don't do it and don't worry about it, there are better ways.

[irldonalb]

Hi All

Thanks for all the comments and feedback.

I’d be really interested to hear the results about the tests mentioned above. However I think we’ll get plausible results and not definitive ones.

Someone mentioned before that no-follow tags on authority sites can have a different behaviour. It’s possible that the no-follow tags on the sites I mentioned above are being ignored. The sites could be doing something stupid, but Google’s ignoring it.

Plus, I think everyone is agreeing it’s not a wise move to do it.

Thanks

[TheMadScientist]

Okay, the preliminary results of the overly simplistic test I started are:

It appears they are using the nofollow url for discovery* at least when one link is nofollowed and one is not, but what I cannot confirm yet is if they counted the link text or PR from either... I used a unique hash for the links and have searched for both and cannot find the page using either, but the page is in the index and shows using a site: command.

If it's the case they use one and not the other as noted above I could see how results could be conflicting and there could be confusion about them, and I could also see how internal and external could be treated differently, which could also lead to some confusion, but it's not something I have tested... The test I did was overly simple, down and dirty, 'let's see what happens', but...

I'll leave it to those who would like to do further testing to do so if they feel the need, because personally I don't really care. :)

* I am doing another test to try and confirm they didn't use something else for discovery, but I'm fairly sure they used the link and will report again if it appears they used another data source some how, because that would be downright unsettling since I don't have the toolbar installed, GA on the site, or open access to stats.

[Sgt_Kickaxe]

I doubt it very much. I think sites are trying to be clever but making problems for themselves instead. This kind of nofollow abuse may be why Matt posted the recent video we're talking about...

I don't see it as abuse to learn how something works and to use it as you see fit.

The fact pagerank and nofollow even exist is what's creating these problems.

[Doogle]

I hope I'm not out of line by using an example but why would yellow pages UK (yell.com) use this on their homepage?

<meta name="robots" content="noindex, nofollow" />

[tedster]

That one's a mystery to me, doogle. And Google is showing the cached page and has it indexed with the exact on-page title. So the noindex part apparently isn't being obeyed.

[Robert Charlton]

I'm not finding the noindex in the source, either on the current page or in the cached page, which is dated July 4. How new is this version of the site?

[tedster]

If this is a brand new change, that might make the difference. I don't see the noindex,nofollow robots meta in the cached version (Jul 4, 2010 20:13:52 GMT) but I do see it in the source code for the live page.