SEOmoz.org recently launched their flagship Linkscape™ tool. If you haven’t heard, welcome back from underneath that rock. I hope the sunlight isn’t too blinding.
Rand Fishkin, founder and CEO, announced the tool at SMX East. I was there. I have the t-shirt to prove it (American Apparel – nice touch, Rand).
Since then I’ve been testing Linkscape, and I’ve come across a flaw that I find glaring.
Here’s a summary of the story behind the tool and how it works:
(click here if you know all this already)
- Search engines, most notably Google, consider links to be a central signal of value or authority on the web (see Google PageRank), and they use links heavily to determine where a web page will rank
- Search engines have a “link graph” of sorts, or a database of all the links they’ve indexed, where they point and their properties
- SEOs to this point have had access to very limited data on this link graph – we only know what search engines deliberately make available
- SEOmoz saw this as a need and saw a way to fulfill it – by building a crawler with the purpose indexing links, storing the information and providing an interface and metrics through which we can view it (for a price)
Here’s what the basic interface looks like once you’re inside an Advanced Report:
There is no doubt – Linkscape provides some incredible information and metrics that we simply have not had before. They’ve crawled some 30 billion pages on the web and will be looking to expand that number in the future. I’m recommending the tool to anybody who wants to know more about their own inbound link profile or that of their competitors.
But there is a plausible misconception some people can have about the data, and I want to point it out.
mozRank is one of the primary metrics used by Linkscape. SEOmoz developed mozRank as an alternative to Google’s PageRank – essentially, mozRank is a measurement of how much “link juice” a page has based on how many links point to it and where those links are published. It follows Google’s assessment of links as the “democratic quality” of the web – each link constituting a vote and an individual site’s “vote power” determined by how many votes point to it.
The trouble with PageRank is that Google has been less than transparent with how they calculate, factor-in and display PageRank. As a result, PageRank, once a focus of great attention, has widely been discounted as just a “little green bar” that provides little actionable data.
SEOmoz seems to put forth its own mozRank as a solid replacement for PageRank and a metric we can actually trust as the basis for educated action. But there’s a problem here, and it stems from how we define “actionable data” when it comes to search.
What makes search engine data actionable?
The main idea behind Linkscape is that SEOs and other web marketers will be able to leverage the data it provides to develop strategies and tactics and improve their search rankings. That’s a simplified take on it, but I think it fits. The question is: does mozRank give us that kind of data?
If we’re going to discuss “taking action” when it comes to SEO, we’ve got to talk about goals. Every action only makes sense through the lens of the goals that have been set. Goals differ from campaign to campaign, but primarily they all tie in to traffic – improving search engine traffic either in volume, quality or both. And if we’re going to talk about traffic, we have to talk about where, specifically, it’s going to come from. In SEO, Google is often the focus – with the lion’s share of the search market Google maintains this should be of no surprise.
So the question follows: how can we leverage mozRank to take the right actions with the goal improving our Google rankings/traffic?
The easy answer is to take mozRank at face value as a more pure and transparent link juice metric than PageRank and, when assessing a link profile, to trust this metric as an indication of how powerful a link from a given page is. The easy answer, however, is almost never the right answer.
Consider the following example from a Linkscape report I ran today. I ran it on an SEO agency (which will remain nameless) website that I saw to have some very competitive positioning in search results. I basically wanted to gain a little more insight into how they’d done it – and Linkscape is a great way to start looking.
By default, once you click over to either the “URL Anchor Text” or “Domain Anchor Text” tab and select an anchor text variation, Linkscape sorts the links by the percentage of mozRank each one passes to the URL/Domain the report was run on. The aim is to display the most important or valuable links for the URL/Domain – the links that are really making a difference in that site’s ability to rank in search engines.
When I was researching the site I mentioned today, I selected one of the most popular variations of anchor text used in their external links that was an exact match for a competitive keyword for which the site ranks at #2 in Google. My goal was to determine what web pages were helping this SEO agency rank so well for such a prominent keyword. The second most powerful link, according to mozRank passed, came from Alive Web Directory (www.alivedirectory.com):
I’ve run across Alive Web Directory before, and to my knowledge this site falls into the category of a “free-for-all” link directory with no significant editorial process – the kind of directory that we’ve all been warned about over the last few years, and the kind of website that Google penalized heavily about a year ago this time in their first serious step going after paid links.
My gut in this case would tell me that Google isn’t going to think much of a link from AliveDirectory.com. But Linkscape is telling me otherwise:
Linkscape has the Alive Web Directory domain with a mozRank (”mR”) of 7.27, the page containing the link with a 6.10 mR, and this link passing 11.28% of this SEO agency website’s 5.74 mR. This is the second most powerful link this website has according to Linkscape.
If I took this data as “actionable,” I might assume that getting a link to my own website from AliveDirectory.com would be a smart move. But I know Google has penalized a ton of free-for-all paid web directories like AliveDirectory.com in the past and that these directories have largely lost their ability to pass PageRank. My next question: what PageRank does AliveDirectory.com have?
They’ve got a nice round PageRank of 0.
This is a site has been up for more than two years and has more than 740,000 inbound links according to Yahoo! Site Explorer.
It should be obvious at this point that the Alive Web Directory was penalized heavily by Google. Indeed, they don’t even rank for their own brand name – a sure sign of a manual penalty on the part of Google’s webspam team.
So what good is mozRank?
This is just one link, I know. I’m hardly providing a wide sampling of these errors. But this is the 2nd most powerful link to an SEO agency website in terms of the percent of mozRank it passed, and it illustrates an important point.
Google’s link data is far more complex than Linkscape’s.
Just last year we witnessed how quickly Google can change the PageRank of a site and take away both its power to rank, even for its own brand/title, and power to pass PageRank to other websites. AliveDirectory.com is a perfect example of this. Every time Google manually penalizes a site this way, Google’s link graph has changed. With one free-for-all directory stripped of its ranking power, the PageRank of an entire network of sites is effected.
Since Linkscape currently does not seem to use Google’s PageRank or other metrics to determine whether a website has been penalized by Google, its entire link graph is thrown into question.
What if we remove the link from Alive Web Directory above from the link graph and mozRank distribution table as Linkscape has it? The entire picture that Linkscape offers as to what links are helping the website rank is altered – and because of a single, prominent link. There are undoubtedly more cases like this in this single link profile. And what about the other websites that Alive Web Directory is linking to? What percentage of mozRank is it passing to them while passing no PageRank in Google’s eyes?
I like Linkscape – there are a ton of great uses for it, and it is already a part of my SEO methodology. However, I’m not sold on mozRank as a reliable metric. If there is no observable correlation between mozRank and PageRank, if we can’t use mozRank to determine what links are truly helping a website rank well in Google or other search engines, is mozRank an actionable metric?
In the end, we’ve got to take what we can get. We’re not going to get our hands on Google’s link graph any time soon. Because there is always going to be a stretch between the assumptions we make in observing search engine ranking practices and what is actually the case, we’ve got to take every tool and metric with a grain of salt.
In the case of Linkscape, the take away here is to understand what mozRank really is: a metric based on the same theory as PageRank but certainly not a representation of PageRank itself.
It’s important to consider what makes data like this actionable or not. Obviously in cases like that of Alive Web Directory, mozRank doesn’t correlate very well to PageRank and Google’s view of what links matter on the web. So we’ve got to ask – what can mozRank actually tell us about the real world of search?
{ 6 comments… read them below or add one }
Great post, and extremely thought provoking. Not having had a chance to use Linkscape very much yet, I would say that with any tool there are pros and cons, and unfortunately this seems to be on of it’s cons.
I’m not sure I agree that it has no actionionable data, again not having used it very much, but lets say you are doing a link review of your competitor. After the report is run you may find a variety of sources for which they have links that you do not. Of course doing your homework on those sites will be important to ensure you aren’t getting links from a “banned” site, but if they are good sites you’ve just found a goldmine of link resources.
Guess it’s time to start digging into LinkScape!
Well, I certainly wouldn’t say that Linkscape offers no actionable data.
I do think that the way mozRank is calculated and displayed doesn’t account for this very important factor. One major link from a site that has been penalized or banned by Google can throw mozRank off by a large degree. That means you’ve got to do a lot more homework on links and where they’re coming from than what Linkscape has to offer, and I think this point renders mozRank as quite a diminished metric from what we might have hoped it would be.
Still, I will continue to use and appreciate Linkscape from here on out. It still offers a better picture of links on the web than search engines do. But I think we need to take it, and especially mozRank, with a great big grain of salt.
Mike – I think you make an excellent point here and the way search engines discount some links that Linkscape does not can obsfucate or provide less than ideal information. I would say that mozRank is still a great metric for SEOs in the know – going and looking at a lot of competitors, you’ll find lots of high value pages (and can sort the links by them) to help get an idea of how juice is flowing and where it comes from. Granted, we’d like to get much better with spam detection and mozTrust is a part of that. We’re also a smaller index (1/3 to 1/5 the size of Google/Yahoo now), but over time, we’ll be getting much bigger and broader.
If there’s any specific things you’d like to see from the tool, don’t hesitate to ping us. We’ve got a ton of feature requests and enhancements we’re working on, but with this much data to draw from, there’s bound to be hundreds if not thousands of great ideas for how we can improve. You can email me personally – rand at seomoz dot org or shoot it to sitesupport at seomoz dot org and the dev team will put it in their pipeline.
@Rand – thanks for stopping by.
No doubt Linkscape offers SEOs the best information on links we’ve ever had. It’s an invaluable addition to my own tool set – and I’m sure it will be for anyone who tries it. Still, this issue was stuck in my craw since I noticed it – mozRank is potentially a solid metric, but “actionable” in my definition has to be tied to how Google and the other search engines actually value links – and in the example above there’s clearly a disconnect.
If nothing else, I wanted to make the case that while Linkscape is a great tool for competitive link research, we still need to cautiously inspect the other signals involved before we take action.
No argument here. It’s definitely a feature I’d like to see as well. The one thing I do like about comparing mozRank vs. PageRank is that it often helps ID who has been penalized or devalued vs. who simply has low PageRank.. For example, when we ran a comparison of 10,000 pages mozRank vs. PageRank, we found that we’re off, on average, by 0.7, which isn’t a huge number, but looking at the top 100-200 where we were off the most significantly, virtually all of those sites had a PR penalty for selling links. Thus, while mozRank might mislead you into thinking that a site passes more juice than it does, using it in combination with PageRank can actually be hugely valuable (and give an extra piece of data you wouldn’t have had previously).
In the future, I think we’ll try to add some sort of notice like “may be penalized” or “may not pass PageRank” – that kind of thing based on the comparison (or based on our own spam/paid link analysis).
That is a great point and idea – if Linkscape displayed an indication of this disconnect in cases like AliveDirectory.com it would be very useful not just for competitive research but to quickly determine whether a site you’re researching has been penalized by Google.