we've been playing with both recently, is there anyone here who has good experience of the Moz pro tools especially Linkscape? what do you think of these two packages against each other? they correspond with each other on the value of some links, but not on others. how much do you trust the metrics? mozRANK etc..?
I actually had a really good chat with Alex from Majestic during my last trip to London for SES. I think that while we're competitors, we're still very optimistic about the space and excited about bringing some third-party metrics to link analysis and competitive analysis. I won't speak to Majestic, just Linkscape, but I will say that the two datasets are crafted very differently - Linkscape is a web crawl, built by bots that spider the web's link structure the same way Google, Yahoo, MSN, etc do. Majestic is built on a distributed software installation to users, so pages and links that are in their index are sent back by users who browse the web (much like Alexa). Because of this, the datasets are very different in how they show information. On Linkscape - our metrics are described in detail here - http://www.seomoz.org/linkscape/help/metrics - but essentially, we have four big ones: 1) mozRank - similar in intuition to Google's PageRank or Yahoo!'s WebRank - it's a score based on an iterative, markov-chain algorithm of the web's link graph on a logarithmic scale. On average, it's off of Google's Toolbar PageRank by about 0.45. This reflects some differences in the way we calculate as well as some differences about application (Google obviously has PageRank penalties on many pages and excludes certain things from its index or canonicalizes them in ways we might not). Comparing mozRank and PageRank is valuable, as well as using mozRank on its own, as it's updated more frequently and shows a greater degree of detail (up to 2 trailing digits, rather than just a single number from 0-10). 2) mozTrust - our interpretation of "TrustRank" - a concept pioneered in a Yahoo! research paper that suggests that using trusted seed sets and crawling out from there can produce a very useful method for sorting spam from legitimate content. Looking at mozRank vs. mozTrust can give you a good idea of sites/pages that have earned lots of links (but from low quality sources) or those that have few links, but from very highly valued sources. 3) Domain Level mozRank (and mozTrust) - It's strongly suspected that all the engines calculate metrics like PageRank/TrustRank on the domain-level graph as well as the page-level graph, and so we've recreated this type of metric across all domains. You can see how big/important/trustworthy a particular subdomain/root domain is vs. others (as opposed to just measuring pages). 4) Link counts - We show an accurate link count (according to our index) of the number of pages, subdomains and root domains that link to a given page/subdomain/root domain. From this, you can get a good idea of link diversity. Hope this helps!
Hey Rand. am honoured to get a personal reply from the CEO of the company, wasnt expecting that on DP on a Friday night havent you got some public speaking to do somewhere? Linkscape in particular does seem like an impressive and powerful tool, but it also seems to highlight certain links that don't look like they'd be doing much if anything at all, as being of major importance. ..and not show certain links that I would have thought would be significant, as doing very much. I suppose my question is, given that we have no real clue as to what G are and aren't doing in their algo, how confident are you that mozRANK really equates to anything meaningful? and would you consider it an attempt at simulation of the real PR figure?
Interesting while I was running a few test sites I found that the tool picked up on the canonicalization issue for larger sites but not so for smaller sites. Very small sample pool though. And one site I tested uses an .htacess redirect to the www version. So I'm interested in how that happens with a webcrawl. Why it didn't just see the redirect and report that. Perhaps it's because it only looks at the first link and doesn't follow it through? PS: Yes I am too tired to read all the doc on the the beta site which probably answer the question. The site tested was "lasvegas.com" which gets a ton of incoming links because it's a type in domain, so their online marketing challenge is pretty low.
SEOibiza - I'm honored you're using the tool and talking about us! In terms of being representative of PageRank, I honestly found it fascinating that across many thousands of pages, the scores mozRank vs. PageRank were only off on average by so little. It tells me that we have a very similar metric, though certainly not exactly the same. I think Ben from SEOmoz is going to be sharing more of that correlation data in the near future on our blog. In terms of predicting rankings, though, we have done some research on that as well - and I actually think that's relatively more interesting than just the PageRank comparison (though being able to spot PR penalized sites is a great feature that mozRank makes much easier). I shared some of that in my presentation at SMX West - and both Domain level mozRank (DmR) and page level mozRank are quite good predictors of ranking success (i.e. if you have more, chances are, you're performing better for a given query). In practical application, the way we use them is to help clients determine good partnerships to form, good domains and sites to get links from or even purchase. We also use it to help in competitive analysis - answering that age old question "why is that site/page ranking above me?" Obviously, PR or mR alone can't answer those, but with so many data points - mozTrust, # of linking domains, anchor text distribution, etc. - we can get a lot closer to those answers. internetmarketingiq - we should be picking up redirects and we try to treat them the same way the engines do (though we'll show how many links and how much mozRank a URL has generated individually, even if it 301s, just so site owners know). If you're finding we've missed something, it could be because our web crawl is older than Google's (our data is, on average, between 30-60 days old),or we could have had a problem. If you think the latter is the case, I'm sure the team would love to know, so feel free to send a report to sitesupport at seomoz.org. I'll also ping our devs about this thread so they can have a look for themselves. Thanks for sharing!
If it is a close-ish simulation of "t3h R34L PR" it's potentially incredibly useful, for example to know whether you're a low or a high "4" etc. look forward to hearing more about this, interesting stuff. I would say this does also appear to hold true on the few specific sites I've looked at and compared performance to mozRANK predicted performance anyway thanks for taking the time to respond, am enjoying playing very much, looks like might have to sign up..
Distributed software yes, links sent back yes, but there is no spyware in the Majestic crawler as in most toolbars who extract user visited links or user favorites/bookmarks. The Majestic crawler get it's to crawl URLs from the majestic server seed URLs and send back URLs found in this crawled seed pages etc and scheduled URLs to recrawl to keep the index fresh so exactly the same as every search engine bot crawl the web. Looks like you never installed the Majestic crawler yourself, the user can even configure what top level domains to crawl. Because this distributed system Majestic crawl rate is higher then the LinkScape crawl rate and also longer running in time what resulted in a much bigger MajesticSeo database then LinkScape and Yahoo. Buy at both system some the same domains and compare the result and with the free Yahoo and you can see the differences yourself, MajesticSeo give a free report of your own domains. Yahoo show only max. 1000 unsorted results and LinkScape 3000 most important links while MajesticSeo show them all. Yahoo do not show anchor text and ranking so is not to compare with this two paid tools. MajesticSeo show daily new backlinks found a feature I like the most, but I like to receive them in a daily RSS feed message so I don't need to login every day also I like to receive a mail or RSS feed message when the database is updated with this daily new found backlinks what happens every 6-8 weeks or a new filter feature is added.
Linkdex - I think that freshness and daily updates is one of Majestic's strongest points. Linkscape only updates every 30-40 days, as our index needs to be rebuilt to calculate the metrics on the link graph (like mozRank, etc). In terms of size - Majestic does report far more URLs & links than Yahoo!, Google (in Webmaster Tools), Linkscape, MSN or any of the other engines on the web. In our testing of Majestic data, many of those appear to be URLs the engines (and Linkscape) would canonicalize (multiple URL parameters, duplicate content, etc.) and about 30-40% of the links reported appear to no longer exist (I think that's because Majestic's crawl can take a long time to refresh a page once it's been seen - although this may just have been for the sample set we browsed). I certainly didn't mean to suggest it's spyware though - like Alexa, it's a user volunteered action to install it.
I have used linkscape and its a good tool but the major problem for introducing features like mozRank and mozTrust made people thinking is this accurate? Even if its accurate does Google gives sames authority which this tool shows? According to their metrics they are close to this but many times information is not reliable, I have run 7-8 reports and some data was useful but some of it was misleading. Also Google's algo changes every now and then and one major change can be threat to this tool. Hope rand will find someway out, as I personally like him and his Whiteboard Friday's.
the bolded part is my question really I suppose. im satisfied that mozRANK is a reasonably close approximation of the basic pagerank idea, and a maybe even a reasonable approximation of ranking potential. and in most cases linkscape agrees with what I would expect to be the best links, but it does also introduce a few randoms that by the usual research methods and intuition you would not think were good links. i wonder, are they, or not? more testing required I suppose..
The question - is mozRank accurate - is an interesting one. The quick answer is yes, at least, it's definitely accurate based on the formula by which it's calculated and on the web link graph it's calculated on. Perhaps the more interesting questions would be: - Is it better/worse than PageRank in predicting ranking positions at Google? - Is it better/worse than PageRank in assessing a raw link popularity metric? - Is it better/worse than PageRank (or other metrics, like Yahoo! link counts or Alexa ranking) in assessing the value of a site/page that you're trying to get a link from? I believe it's our responsibility to do a good job answering all these questions with mathematically rigorous data. We've done some internal projects around this already, and my hope is that in the next month or two, there will be some statistically significant (and compelling) information published via our blog to share. We definitely want to make sure we do this correctly and on broad sample swaths so we can feel very confident about the accuracy. deadlychaos25 - If you have any examples at all where you felt our data was inaccurate or misleading, we'd love to get those. You can email sitesupport at seomoz.org or just leave a comment here or on our blog on Linkscape feedback. Some common sources of inaccuracy are around index size (we somtimes don't show every link you were expecting to see because our index isn't quite as robust as Google/Yahoo!/MSN) and freshness (our data at the moment is 2-3 months old, though we're working to close that gap with every index update).
Stuff getting more interesting here, I agree your tool is more better than PageRank value shown on toolbar. I will mail you some issues about your tool. I am excited to see statistical data as it will really help us determine to check how much we can rely on this tool. Cheers to SEOmoz and Rand (how about buying some new clothes for WBF?).
Just FYI - We updated Linkscape's index last night, so the data you see in there should now be much fresher than last week
I can't speak to Majestic, but here's my opinion of Linkscape-- it is the most robust tool for examining and filtering backlinks available. I think that's where the real benefit is. It gives you a ton of other info with their proprietary metrics, but I'm not exactly sure what to do with all of it.
nope, they get updated automatically. Once you run the report, it gets saved in your account so that you can go through from them afterwards. So just login and see new updated data.