I noticed something interesting today. I was playing with some SEO tools to understand how Google ranks sites a bit better and decided to check how many inbound links Google thinks this site has. I fired up Google and tried:
Google reported a grand total of 248 websites that link to my home page. Now, I don’t think of myself as being vain, but I remember back in the days when my PageRank was 5 that I had tens of thousands of inbound links. So I popped off to ask other search engines what inbound links they though I had:
Well, Yahoo thinks I have over 33,000 links spread over almost 8000 pages.
Now, I suspect this might have something to do with relevance. It’s possible to ask Google what websites it thinks are related to you:
gives a list of websites that Google thinks are related to this one. Which means Google has the ability to figure out what box to put your website in. My theory is that the drop in incoming links could come about from Google only caring about inbound links from related websites. So, if you had a website that lets you search for videographers for example, an inbound link from a website selling acne treatments might not count. I’m not saying this is a bad thing for the searching public as it increases the relevance of the searches. However it does mean that using Page Rank to gauge the value of a website is probably the wrong metric to use.
This theory could explain some things I considered strange on my website. For example, my Comment Email Responder page has a PageRank of 0, even though there are hundreds of bloggers out there who wax lyrical about the plugin. But, if Google thinks the page is about PHP development (which is what it really is), and all the bloggers talking about it don’t run blogs about PHP development (which I’m pretty sure most of them don’t), then Google wouldn’t count those links. What do you think? Does the theory hold water?
The bottom line is that I believe PageRank is really just an internal measure that Google uses to figure out relative relationships between pages in a certain theme and is useless for comparing one webpage to another. Would love to hear your feedback
Are you using Google’s Tools for Webmasters?
That’ll give you more accurate info than link:example.com, it’ll tell you if there are problems with the site, list what position you’re likely to list in certain search queries, etc. It’s pretty damn handy.
Good questions Owen.
It’s been known for some time that Google does not show you all of the backlinks they have indexed for your site. I’m no expert on the topic, but most of the explanations I’ve seen suggest it’s largely an attempt by Google to complicate SEO gaming.
However, unless we believe they purposely inject randomness, there has to be some logic behind what backlinks to report/exclude. Your theory about relevance is as good as any I’ve heard. An interesting experiment would be to select 20+ links that appear in GOOG and Yahoo, and 20+ links that only appear in Yahoo — then compare those sites/links to look for any common themes for inclusion/exclusion.
Someone may have already done this, but it’s probably the type of sampling that might lend some weight (or not) to your theory. Good linkbait also…;-)
@ManxStef: Yeah, I like the Webmaster tools, been following all the guidelines, submitting Sitemaps, pinging services etc, so was quite surprised to see a sum total of 248 websites linking to my front page. I can probably list more than that off the top of my head that actually have a link.
Interesting though …
@Dan: Thanks for your comment mate. It’s interesting to see how large the online advertising market has grown in the absolute absence of any standard metrics across different platform. It’s almost like building a bridge without having a standard way of measuring the length of all the different bits.
Who knows, maybe one day all this will change …
I don’t think pagerank pays attention to relevance of links to your site. However, Google search results definitely do (it is a huge factor). I think pagerank is based just on number of links and authority passed with link (higher value pages with same number of links pass more authority). It is then further confused by Google’s published pagerank which can be reduced (though it seems likely this does not effect search rankings (which is what really matters). I recently posted on Google PageRank Update
Can someone please tell me why my website http://www.spiralteck.com page ranks keeps alternating between pr3 and pr4
I’ve seen this happen to several of my sites all across the board, PR dropped for all of them, but SERPs haven’t changed much.
However, the PR drop for my blog also resulted in PR being scattered to dozens of pages that didn’t have PR assigned to them previously.
So, it’s a weird thing. Google webmaster tools shows tons of “errors” for http://danielmcgonagle.name but it’s the same set up as my other blogs that were set up the same way and the other ones don’t show errors in G webmaster tools.
There are a lot of SEO Tools out there with templatized information and some of it is outright plain wrong, and doesn’t take into consideration the SEO strategy of sites being analyzed.
For instance, if your goal is to get high rankings for pages on site and not just for main KWs in site title or whatever, then the SEO tools don’t know that and can only offer limited “help” based on what they ass-u-me you’re trying to do, i.e. get ranked just for main KWs.
I admit that PR is a nice little green symbol that lets us think we’re oding well with our sites’ SEO, but it is all about rankings and SERPs, no?
BTW, if anyone cares to analyze my site, YSM shows 1000s of inbound links, Google Webmaster shows less, and the numbers vary greatly with all the backlink checker tools I’ve used.
IMHO, if your SERPs aren’t affected, don’t worry about it.
I’ve also noticed that Google picks up blog comments as “indexed” backlinks they choose to show, so maybe that’s something that is still a good strategy, (blog comments for backlinks and relevancy etc…)
I hope this helps someone a bit…
It’s all about the SERPs, not little green symbols