|Links to photos draining PR?|
I was watching a bunch of Matt Cutt's old youtube videos over the weekend and at one point I swear I heard him respond to a question about whether links to photos carry PR, to the affirmative. On a large number of our pages we have say 25 paragraphs of text accompanied by an average of 10 .jpg photos in thumb-print size, which in turn have an HREF link to a full-screen-size .jpg of the same photo. I think this makes total sense as far as visitor experience and useability (a biggie with Google lately) and reduces bandwidth as it allows visitors to only select those that they really want/need to see larger versions of. As a result a lot of our photos are somewhat well ranked in G's image search, but we could do without the traffic from them as they tend not to monetize at all.
However, if what Matt says is true, in the long run, it would seem that this would leak a lot of PR as there is no way to LINK BACK from a .jpg and pass on the PR. basically the PR hits a dead-end if I understood the question and answer. Using a nofollow on the link to the .jpg has no effect since the consensus is that nofollows still leak PR. I imagine one could create a .htm page which had nothing but an IMG tag for every larger photo and a single link back to the home page or other appropriate page. But what would Google think about a huge number of such "empty" pages? Even with a title, description and at most a one line description I think these would be considered shallow content, but at least they would pass back PR.
Is there a standard method or any other suggestions on how to either not pass PR to the large photos or to not set off bells at G for 1000's of shallow pages, while retaining as much PR as possible and focusing it on to the text pages rather than the photos?
A link to a fullscreen.jpg image does not pass PR. Could be the case if you link a .html page with an image as anchor.
Honestly, I wouldn't worry about this at all -- I can't imagine that it would be of any harm.
The shadowbox content might visually pop up "on the same page", but that pop-up box will be indexed as a separate URL.
I've recently met that problem on a shopping site where the product reviews popped up on a shadowbox effect on the product description page. The reviews themselves then also appeared as a separate page in the SERPs. Visiting any of the review pages from the SERPs was problematical as there was no navigation back to the product page or indeed to any place on the site.
That was solved by adding a tiny navigation bar to the reviews page.
There were other challenges to ensure that products with no reviews didn't expose a multitude of blank pages. Likewise for the 'write a review' pages, those contain only a form and had to be excluded from indexing from the start.
A link to a fullsize .jpg certainly passes pagerank to that image. In fact, that is one of the absolute best ways to get images to rank in image search. Instead of putting the large image in the page with a img tag, use a thumbnail in the page and link to the larger version. This works because 1) Google passes pagerank to the image when it is a link but 2) Still considers the image to be part of the page and 3) bigger images just rank better and you can link to really huge ones if you want.
On the other hand, I really doubt that this hurts your site. Google did a lot of work a few years ago to redistribute lost pagerank from your site back to your important pages. There are lots of links that just drop pagerank on the floor: 1) nofollowed links 2) Links to pages in robots.txt 3) Links to images
When I have removed these types of links from my sites, it makes no difference.
There are better things to worry about
Not necessarily, if you are talking 50-90% of internal links go to .jpgs
Hadn't realized larger ones ranking better, but now that you mention it I think I would have to agree. Some of our full page ones rank quite high.
[edited by: MikeNoLastName at 1:10 am (utc) on Jun 5, 2012]
You could block the free-standing images from Image Search-- which never shows them at full size anyway-- and allow only the thumbnail versions. That way people could never land on your big images "cold", but only via the proper link.
You may or may not want to block non-referer or cookieless humans from seeing the picture. Fine if they've got it bookmarked and started out visiting it by official means, but otherwise iffy. Do you have much of a hotlink problem? So far, that's been my sole reason for blocking areas from Image Search.
Oh. Oops. You were asking about g###. I was thinking of humans. Never mind then.
I really want to avoid scripts if possible as it never seems predictable how the SE's are going to handle them from one week to another.
That's an interesting idea lucy24 blocking the large ones. Of course you'd still be losing link juice either way since they are still counted as a link off the page.
Not really so much of a hotlink issue (bloggers and our competitors just outright scrape them and make us send them DMCA notices :), but more we get a lot of hits on the large images and then they bounce never visiting another page or even the page it was linked from, skewing our bounce stats also. I assume G is using the ALT tag and title to rank them.
Just to clarify to all, I'm talking about a simple link on a .html page like:
blah, blah, blah
<A HREF="example.jpg"><IMG SRC="example-th.jpg" ALT="larger photo of example"></A>
blah, blah, blah
Matt Cutts unequivocally said that linkjuice is passed down image links. So people hot linking to your images are actually doing you a favour. On the flip side yes if you heavily into pr sculpting on your site then this will pose a problem. personally I don't worry about it to much.
Not necessarily, if you are talking 50-90% of internal links go to .jpgs]
2 years ago I received a new client with PR0 penalty. After a while I found out that almost all images were "hijacked" from another site. This was done unintentionally (he was a distributor who simply wanted to use images from the main catalog) and still it took us almost 2 month to remove the penalty.
It`s always better to use your own assets.
BTW: Penguin is much harsher on hot-linking.
Nothing is "too small" that`s why it`s called "optimization".
|I heard him respond to a question about whether links to photos carry PR |
Yes, image links pass PR just like any other link.
This is the video titled:
"Does PageRank flow through image links?"
The answer is a big emphatic "YES!".
If you don't like that answer, take it up with Matt.
I'm just the messenger ;)
I've noticed a problem with rankings of html pages carrying too many image thumbnails. I've posted about it here [webmasterworld.com...] thinking that the problem was because of too many alt tags with the related keywords inserted. Now, there's also possibility that the problem is due to simply too many thumbnails. It appears that it doesn't matter if the images are opened directly as .jpg, via shadowbox, or as an .html image attachment page (wordpress' feature).
Yep, that's the one, thanks
"Matt Cutts unequivocally said that linkjuice is passed down image links."
One of my points though is, does Matt know what the questioner was calling an "image link" when answering that way? What exactly is the definition of "image link" in the context of the question? Is it a link to a .htm using an image, or is it a link TO an image?
For example is he referring to the example I gave above (linking to an IMAGE (i.e. .jpg) which will go to a new browser window, with either an image (or plain anchor text for that matter) passing rank to the linked .jpg) or did he mean the following: <A HREF="example.htm"><IMG SRC="example.jpg" ALT="example"></A> where JUST an image links to a .htm (passing PR to a .htm)... or both?
Every html reference source on the net that I could find designates a "link" as ALWAYS using an <A> tag, thus by that definition JUST the <IMG SRC="example.jpg" ALT="example"> alone is NOT considered an image link. Thus image "hotlinks" aka "Inline linking, leeching, piggy-backing, direct linking, offsite image grabs, bandwidth theft, etc" (as defined on wiki-p) are technically not "links" at all and are NOT covered by the technical definition of image link.
"So people hot linking to your images are actually doing you a favour."
Sure we wouldn't mind that, especially if they indeed passed PR which can't technically be inferred by Matt's response. We don't mind others using them if we get credit. We used to look for that via our logs and just change the picture a bit to include our URL to get the free advertising. But nowadays they just outright copy them, maybe modify them slightly, and put them on their own sites and link them from there.
"It`s always better to use your own assets."
Many of ours are original photos or scans, others are PR photos that we are granted full permission to modify and use by the photographer or owner and we post photo credits when requested, but they are always hosted on our own server.
danijelzi: interesting thread, but not the same scenario.
"thinking that the problem was because of too many alt tags with the related keywords inserted.... as an .html image attachment page (wordpress' feature)."
Or is the issue that G doesn't like wordpress .html image attachment pages, considering them shallow content?
The point of this question was the lack of ability to pass on PR from a linked .jpg image and thus how to avoid passing PR to them. If you have page a.htm with link juice=1 it could link to pages 1.htm through 10.htm and split it's outgoing PR among them. They then each have a .10 juice coming in and say .1x90%=.09 to link out to other pages. They could link it back to a.htm or the home page, thus recycling that PR. However If a.htm is ALSO linking 10 .jpgs (via HREFs not IMGs) each of 1.htm-10.htm only get .05 to pass on but the other .5 is lost forever because the .jpgs can't pass on their collective .45 (.05 each x 90%) because you can't out-link from a .jpg. So that would imply that thumbnail linking like this is undesirable.
Doing a little more research the other night I noticed that when G includes one of our linked LARGE images in their image results, they generally include the page with the THUMBNAIL that links to it as the shadow SOURCE page behind it. Thus you DO get some traffic to the .html (if the person clicks to close the G image window - hmm, does G count themselves accessing the page to display behind the image as a hit/bounce in WMT if the person doesn't access the html page?) from the photo image, but that does not indicate you're not losing valuable PR via the .jpg dead-end.
If you are using Wordpress and link thumbnails to attachment pages (.html or such) instead of directly .jpg, WP will create a page like yoursite.com/blue-widget.html/blue-widget-front-view, with the image and maybe its caption as the only content. The problem here is that yoursite.com/blue-widget.html/blue-widget-front-view often outranks yoursite.com/blue-widget.html on Google and users land on a page with only image on it, instead on the full article and this can cause high bounce rates. I don't know the reason why this is happening, but maybe the part of the problem is the url structure, where yoursite.com/blue-widget.html/blue-widget-front-view appears to be better optimized than just blue-widget.html. If you have a lot of attachment pages, this likely drains PR a lot and creates duplicate and thin content. I noticed this problem back in 2010, but maybe they started penalizing this with Panda in 2011.
Correction for "(.html or such)". Wordpress actually doesn't put extensions on ends of attachment page urls.
The video where Matt Cutts states that PageRank "flows" through image links: How can PR flow through image links when images can't actually link to anything? If he stated that PageRank flowed TO images, then we would have something conclusive.
Additionally, I have yet to come across an image with Toolbar PR.
Since I link to a lot of higher resolution photos purely for my users, I'm thinking I might start showing links to high resolution photos only if the user has the user agent of one of the major browsers, thus excluding Googlebot.
|How can PR flow through image links when images can't actually link to anything? |
An image as a link:
<img border="0" alt="W3Schools" src="logo_w3s.gif" width="100" height="100" />
That's a thumbnail linking to a page. If the link says
<a href = "/images/greatbigpicture.jpg" target = "_blank">
<img src = blahblah...></a>
then once you're on the great big picture, you can't go anywhere.
I now realize that I've noticed the same phenomenon as Mike: big picture in Image Search, combined with the gallery page (including all thumbnails) it belongs to. That is, for Image Search purposes there doesn't seem to be any difference between <a href> and <img src>. In logs it tends to come through as two copies of the gallery page, with the blowup sandwiched in between. Not sure what user behavior this reflects; it's not an artifact of the search itself, because there are several seconds between.
Some robots definitely make the distinction, though. I've seen a good number of those single-attack blitzes where they follow every <a href> regardless of what's at the other end.
PageRank is about pages NOT images.
>> An image as a link: <a href="http://www.w3schools.com"><img border="0" alt="W3Schools" src="logo_w3s.gif" width="100" height="100" /></a>
Hrm. I misunderstood then. I thought Matt Cutts was saying PR flowed to images.
[edited by: Andem at 6:56 pm (utc) on Jun 6, 2012]
This shows how PageRank flows to [w3schools.com...] (a page).
<a href="http://www.w3schools.com"><img border="0" alt="W3Schools" src="logo_w3s.gif" width="100" height="100" /></a>
From what I understand in the first part of this thread, the discussion is about whether PageRank flowed to image files. From my experience, it doesn't but the pages linking to the image files can have an effect to how they rank (SERP, not PR).
What we need is some A/B testing to figure out whether linking to images can be a drain on on-site PageRank.
If you had a PR5 (Page A) page linking to 4 JPEG files and 1 other page (Page B), would 100% of the PageRank that is passable flow only to Page B, or would those JPEG files eat up some up the PageRank which would have normally flowed to Page B (without the JPEG links)?
|From what I understand in the first part of this thread, the discussion is about whether PageRank flowed to image files. From my experience, it doesn't |
This should be relatively easy to check for any site that links to an image URL rather than an HTML page. Just check the toolbar PR for the image's URL, right? None of my sites linke to image URLs, so I have nothing that I can easily check.
If there is a PR value, then PR is being voted through that link. This means that other links on the parent HTML page (for instance, navigation) are voting a smaller portion of the total available PR to their target pages.
However, all that is based on the original PR model, and that model is long gone. What we now have has been called the "reasonable surfer" model and PR transfer is heavily weighted toward the links the visitor is more likely to actually use. In fact, given the data Google collects from Chrome users, I wouldn't be surprised if this weighting is using real click data for at leasts part of the assigned weight.
However, since image URLs only show up in Image Search and not organic rankings, it does seem that something else must also be involved.