|Self-referencing links: PR drain, UX nuisance or harmless?|
I've looked at Analytics In-Page reports today and found an insane amount of clicks from a page going to the same page which the CMS links by default from within the breadcrumbs section, pagination and then several more times in the comments section, this time with named anchors attached to the end of the URL. It looks like on average 50% of clicks from a page go right back to the same page, and sometimes it's as high as 70%+!
So, this got me thinking: are these self-referencing links harmful in any way? I can see how they may be a nuisance to the user if they click and simply reload the same page. I'm not sure exactly why do they click on that link in the first place - it's the same anchor as <h1> right above it, only linked back to the same page.
But, the actual user experience aside, what would this do to the way Google sees my pages if using UX as ranking metric: do I get a "credit" for the second visit to the same page? Do I get an "infraction" for shorter stay on the page if, presumably, the entire stay would they get divided between two page loads? Do they even consider those as separate visits given that they are only a few seconds apart most of the time? Do these links unnecessarily dilute PR going to other pages?
Some of those self-referencing links are unavoidable: the comment permalinks are actually different URIs since they have the anchor at the end, although they lead to the same page. But I can adjust the CMS to remove those self-referencing links from breadcrumbs and navigation.
What do you guys think is the best way to handle those self-referencing links? Should I even bother?
Thanks for all your input!
fwiw: This user can't stand 'em. No matter how common they are, you simply don't expect a link to refer back to the same page you're already on, so you're left feeling like an idiot. When I finally caved in and learned three words of php, it was for this specific reason: so I could auto-generate navigation footers without including links to the current page.
Practical question: Does GA distinguish between links and reloads? I see a fair number of reloads in piwik, and can never figure out why they count them at all. Isn't there some way an analytics program can tell if you've got the kind of page whose content might really change in ten seconds? (Forums, say. Or manually loaded webcams.)
I would try a (real) click heatmap to see if the visitors are actually clicking the link.
I've done some testing on this. Links to the same page appear to be ignored in Google's pagerank algorithm.
Remove them if they are hurting the user experience.
My rule of thumb is never link to the current page from the current page. It's highly confusing.
In terms of your specifics, I don't like terms like "PR drain" at all. But I would say it is both a UX nuisance and a poor means to distribute your internal link value.
@Lucy - there are certainly ways analytics can tell. But most will record another page load in the same way as clicking a link.
|Links to the same page appear to be ignored in Google's pagerank algorithm. |
Does this mean anchor links (those with the # sign linking, for example, from the top part of a page to a specific section on the same page) are not counted by Google?
I am also wondering, if you have a lot of anchor links on a page and they all contain a keyword phrase, could this be seen as over optimization or are they ignored?
Thank you for your input, guys. I am going to go over the parts of the code that create those links and see if I can reduce the amount of them. I hope there won't be an over/under/proto/quasi/pseudo/extra optimization penalty in store for me just for removing a couple of my own self-referencing links.
As a general note, however, I think Google Analytics grossly over-reports the amount of clicks on those links. For one thing, they sum up all links to the same URL and show the resulting percentage for each of the links pointing to that URL. That by default makes it the most used link - there are not many (perhaps not even one) other links with URLs that repeat on a page, only those self-referencing ones. Also, they bunch up together straight links for a URL and anchored links to sections of the same URL, so that's I think how the insanely high CTR percentages come about.
Even so, there's a huge difference between 50% CTR for the self-referencing link and something like 5% for the next most clicked link. So, I have to assume there's enough people clicking on those links and in the end simply reloading the page.
@gouri: my most prominently positioned self-referencing link has the same exact text as the H1 and title tags. I'm going to remove it and I'm bracing myself for the wrath of Google - I'll post results here in a few weeks.
Links with # in them are treated differently than links without by the pagerank algorithm.
Re self referencing links - ewww. Hard to see that as having any kind of user benefit, so I'd run away fast from that one.
Links with # in them may be treated differently than links without them by the pagerank algorithm, but can they contribute to over optimization?
Is the anchor text of a link when the url it is going to has a # in it also looked at it in terms of exact match, partial match, etc., and evaluated for keyword stuffing? You could have many of these types of links on a page.
Some self referencing links are pretty much unavoidable for most sites... specifically in your navigation (top navigation and left navigations if using inverted "L" type site architecture).
In most main navigations, you'll highlight the navigation entry that was clicked on to arrive at a page. Very few sites actually change a navigation menu item from a link to simply a label once it's clicked on. Typically they remain a link with a different backgrownd color.
As long as the self-referencing link is part of the page template (header, navigation, footer) it likely has no inpact on rankings. Sites have been built this way since the beginning of time. It's standard practice for most every CMS theme and non-CMS web site on the web. So I don't see the practice negatively affecting rankings.
But if you're placing self-referencing links in the main content portion of the page (within the article on the page for example), that could very well be a red flag.
< moved from another location >
I am wondering if anchor links (those with # sign in them) on a website are counted as part of the total number of internal links on a website.
Suppose you have 10 internal links on a website that go to other pages on the site and you have 5 anchor links, and you want to determine the percentages of exact-match anchor text and partial-match anchor text.
You would need to know the total number of links. Is the total number 10 or 15?
[edited by: Robert_Charlton at 9:26 pm (utc) on Mar 6, 2013]
|I am wondering if anchor links (those with # sign in them) on a website are counted as part of the total number of internal links on a website. |
I don't know if this answers your question, but if you look at the Anchor Text area in wmt, there are three different things thrown together in the same "How your data is linked" list. Anchor text from outside; linking text from your own navigation; linking text from intenal # links. You can't tell which is which unless you already know.
It would be interesting to know whether there's any difference between external links to a page, and external links to a named # anchor. But I don't think there is any way to find out unless they see fit to put out a video that says so.
Wouldn't you think they could at least color code the different categories? It would be phenomenally useful to see at a glance if the text other people are using to link to you is consistently different from the text you use to link to yourself. ("I'm the world's leading authority on blue widgets? Dang. I thought this page was about small furry widgets of all colors.")
|I don't know if this answers your question, but if you look at the Anchor Text area in wmt, there are three different things thrown together in the same "How your data is linked" list. Anchor text from outside; linking text from your own navigation; linking text from intenal # links. You can't tell which is which unless you already know. |
I didn't know that internal # links are shown in the "How your data is linked" list. I think when you say internal # links, you mean links on a page linking to a specific section of the same page and not links going from one page to another, but I just wanted to make sure?
My internal # links, I believe, are not showing. Could this be because in this list, only the most popular links are shown, and the internal # links that I have only appear on the site once? They don't repeat a couple of times. They all have different anchor texts.
Could the fact that the 3 types of links are all in the list mean that the # links are counted as internal links and, as a result, the answer to my question would be 15?
Also, this could also suggest that the navigation links are also counted? Some have mentioned that the search engines differentiate links based on where they are located (e.g., navigation, body text) and maybe they do, but maybe internal links from different parts of a page are looked at.
|I think when you say internal # links, you mean links on a page linking to a specific section of the same page and not links going from one page to another, but I just wanted to make sure? |
Oh, good point. I don't have a way to check systematically. The few that I looked up individually turned out to be cross-page links. That is, links in the form "/directory/pagename.html#anchor".
But as you say, the list cuts out after some number of items, so same-page links generally wouldn't show up since there's only one of them. Unless, in my case, they happen to come at the very beginning of the alphabet. I didn't see any that were unambiguous.
Navigation links are definitely listed. But then, I don't know how good search engines are at identifying navigation sections. If it's a widely used CMS that comes with something like "id = 'navheader'" then maybe, but otherwise how can they tell?
|Navigation links are definitely listed. But then, I don't know how good search engines are at identifying navigation sections. If it's a widely used CMS that comes with something like "id = 'navheader'" then maybe, but otherwise how can they tell? |
Maybe the links that are shown are based on the quantity, and not so much on where on the page they are located.
Would you say that same-page anchor links are counted as an internal link, just based on your general opinion of how links are looked at?
No idea. This is where I step back and let the grownups take over ;)
technically the url of the document stops before the #.
this is the url google indexes.
the part after the hash mark/pound sign is a "fragment identifier" and the legacy term "anchor" is based on now-deprecated markup.
google will sometimes use the url-with-fragment-identifier for a "Jump to" link in the SERP snippet.
for example, a search for many of the apache directives will show the url of the relevant document and an additional link within the snippet the specific document fragment that applies to your search.
apache rewriterule - Google Search:
if i had to guess they include the "Jump to" link based on a preponderance of usage in inbound links.
discussion here - Seeing more Google jump-to options now:
google will also sometimes show a row of "sitelinks" that are referring to fragment identifiers in the indexed document.
RewriteCond - Google Search:
from my observations this is typically occurs in conjunction with a well-constructed navigation menu for on-page content.
they first started doing this type of sitelinks for wikipedia and it was discussed here several years ago.
Page Fragment Navigation in Mini Sitelinks - and snippet!:
|google will also sometimes show a row of "sitelinks" that are referring to fragment identifiers in the indexed document. |
Could this mean that anchor # links within the same page are considered important by Google and these anchor texts should be counted when totaling internal links on a site and calculating percentages of exact- and partial-match anchor texts?
If they are shown in the SERP, maybe they are considered important.
|if i had to guess they include the "Jump to" link based on a preponderance of usage in inbound links. |
Can't be the only variable, though. Admittedly it's hard to tell unless the request is from a search engine or similar, where you see the whole thing including fragment # in the Referer box. (Or in analytics. Piwik tracks fragments, so I assume GA does too.)
But I've got search-engine fragment visits that I'm tolerably sure haven't been linked in that form from anywhere on the outside. My own best guess is that they look for headers or do I mean headings? in the vicinity of the search term. If your target is clustered around an h3 near the bottom of the page, and that h3 has a name and/or id of its own, that's where the search engine sends you.
What I'm describing happens to the first fragment search that I ever noticed, a year or two back. But the one I currently see most often is also associated with an h3-- and the name of the anchor is one of the search terms.
:: pause for belated "D'oh!" moment ::
I just realized I've got one bit of local-anchor text that's common enough that it would show up in gwt if they listed it: The ubiquitous "back to top" that occurs several times on most pages.
:: quick detour to check ::
Nope, not there. That suggests they list cross-page anchor text but not same-page text. But there's no difference-- color coding or similar-- between your own linking text, whether full-page or fragment, and outside linking text.
Google? You listening? That would be useful.