Big Data, Big Problems: 4 Major Link Indexes Compared

Posted by russangular

Given this blog’s readership, chances are good you will spend some time this week looking at backlinks in one of the growing number of link data tools. We know backlinks continue to be one of, if not the most important
parts of Google’s ranking algorithm. We tend to take these link data sets at face value, though, in part because they are all we have. But when your rankings are on the line, is there a better way to get at which data set is the best? How should we go
about assessing these different link indexes like
Moz,
Majestic, Ahrefs and SEMrush for quality? Historically, there have been 4 common approaches to this question of index quality…

  • Breadth: We might choose to look at the number of linking root domains any given service reports. We know
    that referring domains correlates strongly with search rankings, so it makes sense to judge a link index by how many unique domains it has
    discovered and indexed.
  • Depth: We also might choose to look at how deep the web has been crawled, looking more at the total number of URLs
    in the index, rather than the diversity of referring domains.
  • Link Overlap: A more sophisticated approach might count the number of links an index has in common with Google Webmaster
    Tools.
  • Freshness: Finally, we might choose to look at the freshness of the index. What percentage of links in the index are
    still live?

There are a number of really good studies (some newer than others) using these techniques that are worth checking out when you get a chance:

  • BuiltVisible analysis of Moz, Majestic, GWT, Ahrefs and Search Metrics
  • SEOBook comparison of Moz, Majestic, Ahrefs, and Ayima
  • MatthewWoodward
    study of Ahrefs, Majestic, Moz, Raven and SEO Spyglass
  • Marketing Signals analysis of Moz, Majestic, Ahrefs, and GWT
  • RankAbove comparison of Moz, Majestic, Ahrefs and Link Research Tools
  • StoneTemple study of Moz and Majestic

While these are all excellent at addressing the methodologies above, there is a particular limitation with all of them. They miss one of the
most important metrics we need to determine the value of a link index: proportional representation to Google’s link graph
. So here at Angular Marketing, we decided to take a closer look.

Proportional representation to Google Search Console data

So, why is it important to determine proportional representation? Many of the most important and valued metrics we use are built on proportional
models. PageRank, MozRank, CitationFlow and Ahrefs Rank are proportional in nature. The score of any one URL in the data set is relative to the
other URLs in the data set. If the data set is biased, the results are biased.

A Visualization

Link graphs are biased by their crawl prioritization. Because there is no full representation of the Internet, every link graph, even Google’s,
is a biased sample of the web. Imagine for a second that the picture below is of the web. Each dot represents a page on the Internet,
and the dots surrounded by green represent a fictitious index by Google of certain sections of the web.

Of course, Google isn’t the only organization that crawls the web. Other organizations like Moz,
Majestic, Ahrefs, and SEMrush
have their own crawl prioritizations which result in different link indexes.

In the example above, you can see different link providers trying to index the web like Google. Link data provider 1 (purple) does a good job
of building a model that is similar to Google. It isn’t very big, but it is proportional. Link data provider 2 (blue) has a much larger index,
and likely has more links in common with Google that link data provider 1, but it is highly disproportional. So, how would we go about measuring
this proportionality? And which data set is the most proportional to Google?

Methodology

The first step is to determine a measurement of relativity for analysis. Google doesn’t give us very much information about their link graph.
All we have is what is in Google Search Console. The best source we can use is referring domain counts. In particular, we want to look at
what we call
referring domain link pairs. A referring domain link pair would be something like ask.com->mlb.com: 9,444 which means
that ask.com links to mlb.com 9,444 times.

Steps

  1. Determine the root linking domain pairs and values to 100+ sites in Google Search Console
  2. Determine the same for Ahrefs, Moz, Majestic Fresh, Majestic Historic, SEMrush
  3. Compare the referring domain link pairs of each data set to Google, assuming a
    Poisson Distribution
  4. Run simulations of each data set’s performance against each other (ie: Moz vs Maj, Ahrefs vs SEMrush, Moz vs SEMrush, et al.)
  5. Analyze the results

Results

When placed head-to-head, there seem to be some clear winners at first glance. In head-to-head, Moz edges out Ahrefs, but across the board, Moz and Ahrefs fare quite evenly. Moz, Ahrefs and SEMrush seem to be far better than Majestic Fresh and Majestic Historic. Is that really the case? And why?

It turns out there is an inversely proportional relationship between index size and proportional relevancy. This might seem counterintuitive,
shouldn’t the bigger indexes be closer to Google? Not Exactly.

What does this mean?

Each organization has to create a crawl prioritization strategy. When you discover millions of links, you have to prioritize which ones you
might crawl next. Google has a crawl prioritization, so does Moz, Majestic, Ahrefs and SEMrush. There are lots of different things you might
choose to prioritize…

  • You might prioritize link discovery. If you want to build a very large index, you could prioritize crawling pages on sites that
    have historically provided new links.
  • You might prioritize content uniqueness. If you want to build a search engine, you might prioritize finding pages that are unlike
    any you have seen before. You could choose to crawl domains that historically provide unique data and little duplicate content.
  • You might prioritize content freshness. If you want to keep your search engine recent, you might prioritize crawling pages that
    change frequently.
  • You might prioritize content value, crawling the most important URLs first based on the number of inbound links to that page.

Chances are, an organization’s crawl priority will blend some of these features, but it’s difficult to design one exactly like Google. Imagine
for a moment that instead of crawling the web, you want to climb a tree. You have to come up with a tree climbing strategy.

  • You decide to climb the longest branch you see at each intersection.
  • One friend of yours decides to climb the first new branch he reaches, regardless of how long it is.
  • Your other friend decides to climb the first new branch she reaches only if she sees another branch coming off of it.

Despite having different climb strategies, everyone chooses the same first branch, and everyone chooses the same second branch. There are only
so many different options early on.

But as the climbers go further and further along, their choices eventually produce differing results. This is exactly the same for web crawlers
like Google, Moz, Majestic, Ahrefs and SEMrush. The bigger the crawl, the more the crawl prioritization will cause disparities. This is not a
deficiency; this is just the nature of the beast. However, we aren’t completely lost. Once we know how index size is related to disparity, we
can make some inferences about how similar a crawl priority may be to Google.

Unfortunately, we have to be careful in our conclusions. We only have a few data points with which to work, so it is very difficult to be
certain regarding this part of the analysis. In particular, it seems strange that Majestic would get better relative to its index size as it grows,
unless Google holds on to old data (which might be an important discovery in and of itself). It is most likely that at this point we can’t make
this level of conclusion.

So what do we do?

Let’s say you have a list of domains or URLs for which you would like to know their relative values. Your process might look something like
this…

  • Check Open Site Explorer to see if all URLs are in their index. If so, you are looking metrics most likely to be proportional to Google’s link graph.
  • If any of the links do not occur in the index, move to Ahrefs and use their Ahrefs ranking if all you need is a single PageRank-like metric.
  • If any of the links are missing from Ahrefs’s index, or you need something related to trust, move on to Majestic Fresh.
  • Finally, use Majestic Historic for (by leaps and bounds) the largest coverage available.

It is important to point out that the likelihood that all the URLs you want to check are in a single index increases as the accuracy of the metric
decreases. Considering the size of Majestic’s data, you can’t ignore them because you are less likely to get null value answers from their data than
the others. If anything rings true, it is that once again it makes sense to get data
from as many sources as possible. You won’t
get the most proportional data without Moz, the broadest data without Majestic, or everything in-between without Ahrefs.

What about SEMrush? They are making progress, but they don’t publish any relative statistics that would be useful in this particular
case. Maybe we can hope to see more from them soon given their already promising index!

Recommendations for the link graphing industry

All we hear about these days is big data; we almost never hear about good data. I know that the teams at Moz,
Majestic, Ahrefs, SEMrush and others are interested in mimicking Google, but I would love to see some organization stand up against the
allure of
more data in favor of better data—data more like Google’s. It could begin with testing various crawl strategies to see if they produce
a result more similar to that of data shared in Google Search Console. Having the most Google-like data is certainly a crown worth winning.

Credits

Thanks to Diana Carter at Angular for assistance with data acquisition and Andrew Cron with statistical analysis. Thanks also to the representatives from Moz, Majestic, Ahrefs, and SEMrush for answering questions about their indices.

Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don’t have time to hunt down but want to read!

Reblogged 4 years ago from tracking.feedpress.it

Simple Steps for Conducting Creative Content Research

Posted by Hannah_Smith

Most frequently, the content we create at Distilled is designed to attract press coverage, social shares, and exposure (and links) on sites our clients’ target audience reads. That’s a tall order.

Over the years we’ve had our hits and misses, and through this we’ve recognised the value of learning about what makes a piece of content successful. Coming up with a great idea is difficult, and it can be tough to figure out where to begin. Today, rather than leaping headlong into brainstorming sessions, we start with creative content research.

What is creative content research?

Creative content research enables you to answer the questions:

“What are websites publishing, and what are people sharing?”

From this, you’ll then have a clearer view on what might be successful for your client.

A few years ago this required quite an amount of work to figure out. Today, happily, it’s much quicker and easier. In this post I’ll share the process and tools we use.

Whoa there… Why do I need to do this?

I think that the value in this sort of activity lies in a couple of directions:

a) You can learn a lot by deconstructing the success of others…

I’ve been taking stuff apart to try to figure out how it works for about as long as I can remember, so applying this process to content research felt pretty natural to me. Perhaps more importantly though, I think that deconstructing content is actually easier when it isn’t your own. You’re not involved, invested, or in love with the piece so viewing it objectively and learning from it is much easier.

b) Your research will give you a clear overview of the competitive landscape…

As soon as a company elects to start creating content, they gain a whole raft of new competitors. In addition to their commercial competitors (i.e. those who offer similar products or services), the company also gains content competitors. For example, if you’re a sports betting company and plan to create content related to the sports events that you’re offering betting markets on; then you’re competing not just with other betting companies, but every other publisher who creates content about these events. That means major news outlets, sports news site, fan sites, etc. To make matters even more complicated, it’s likely that you’ll actually be seeking coverage from those same content competitors. As such, you need to understand what’s already being created in the space before creating content of your own.

c) You’re giving yourself the data to create a more compelling pitch…

At some point you’re going to need to pitch your ideas to your client (or your boss if you’re working in-house). At Distilled, we’ve found that getting ideas signed off can be really tough. Ultimately, a great idea is worthless if we can’t persuade our client to give us the green light. This research can be used to make a more compelling case to your client and get those ideas signed off. (Incidentally, if getting ideas signed off is proving to be an issue you might find this framework for pitching creative ideas useful).

Where to start

Good ideas start with a good brief, however it can be tough to pin clients down to get answers to a long list of questions.

As a minimum you’ll need to know the following:

  • Who are they looking to target?
    • Age, sex, demographic
    • What’s their core focus? What do they care about? What problems are they looking to solve?
    • Who influences them?
    • What else are they interested in?
    • Where do they shop and which brands do they buy?
    • What do they read?
    • What do they watch on TV?
    • Where do they spend their time online?
  • Where do they want to get coverage?
    • We typically ask our clients to give us a wishlist of 10 or so sites they’d love to get coverage on
  • Which topics are they comfortable covering?
    • This question is often the toughest, particularly if a client hasn’t created content specifically for links and shares before. Often clients are uncomfortable about drifting too far away from their core business—for example, if they sell insurance, they’ll typically say that they really want to create a piece of content about insurance. Whilst this is understandable from the clients’ perspective it can severely limit their chances of success. It’s definitely worth offering up a gentle challenge at this stage—I’ll often cite Red Bull, who are a great example of a company who create content based on what their consumers love, not what they sell (i.e. Red Bull sell soft drinks, but create content about extreme sports because that’s the sort of content their audience love to consume). It’s worth planting this idea early, but don’t get dragged into a fierce debate at this stage—you’ll be able to make a far more compelling argument once you’ve done your research and are pitching concrete ideas.

Processes, useful tools and sites

Now you have your brief, it’s time to begin your research.

Given that we’re looking to uncover “what websites are publishing and what’s being shared,” It won’t surprise you to learn that I pay particular attention to pieces of content and the coverage they receive. For each piece that I think is interesting I’ll note down the following:

  • The title/headline
  • A link to the coverage (and to the original piece if applicable)
  • How many social shares the coverage earned (and the original piece earned)
  • The number of linking root domains the original piece earned
  • Some notes about the piece itself: why it’s interesting, why I think it got shares/coverage
  • Any gaps in the content, whether or not it’s been executed well
  • How we might do something similar (if applicable)

Whilst I’m doing this I’ll also make a note of specific sites I see being frequently shared (I tend to check these out separately later on), any interesting bits of research (particularly if I think there might be an opportunity to do something different with the data), interesting threads on forums etc.

When it comes to kicking off your research, you can start wherever you like, but I’d recommend that you cover off each of the areas below:

What does your target audience share?

Whilst this activity might not uncover specific pieces of successful content, it’s a great way of getting a clearer understanding of your target audience, and getting a handle on the sites they read and the topics which interest them.

  • Review social profiles / feeds
    • If the company you’re working for has a Facebook page, it shouldn’t be too difficult to find some people who’ve liked the company page and have a public profile. It’s even easier on Twitter where most profiles are public. Whilst this won’t give you quantitative data, it does put a human face to your audience data and gives you a feel for what these people care about and share. In addition to uncovering specific pieces of content, this can also provide inspiration in terms of other sites you might want to investigate further and ideas for topics you might want to explore.
  • Demographics Pro
    • This service infers demographic data from your clients’ Twitter followers. I find it particularly useful if the client doesn’t know too much about their audience. In addition to demographic data, you get a breakdown of professions, interests, brand affiliations, and the other Twitter accounts they follow and who they’re most influenced by. This is a paid-for service, but there are pay-as-you-go options in addition to pay monthly plans.

Finding successful pieces of content on specific sites

If you’ve a list of sites you know your target audience read, and/or you know your client wants to get coverage on, there are a bunch of ways you can uncover interesting content:

  • Using your link research tool of choice (e.g. Open Site Explorer, Majestic, ahrefs) you can run a domain level report to see which pages have attracted the most links. This can also be useful if you want to check out commercial competitors to see which pieces of content they’ve created have attracted the most links.
  • There are also tools which enable you to uncover the most shared content on individual sites. You can use Buzzsumo to run content analysis reports on individual domains which provide data on average social shares per post, social shares by network, and social shares by content type.
  • If you just want to see the most shared content for a given domain you can run a simple search on Buzzsumo using the domain; and there’s also the option to refine by topic. For example a search like [guardian.com big data] will return the most shared content on the Guardian related to big data. You can also run similar reports using ahrefs’ Content Explorer tool.

Both Buzzsumo and ahrefs are paid tools, but both offer free trials. If you need to explore the most shared content without using a paid tool, there are other alternatives. Check out Social Crawlytics which will crawl domains and return social share data, or alternatively, you can crawl a site (or section of a site) and then run the URLs through SharedCount‘s bulk upload feature.

Finding successful pieces of content by topic

When searching by topic, I find it best to begin with a broad search and then drill down into more specific areas. For example, if I had a client in the financial services space, I’d start out looking at a broad topic like “money” rather than shooting straight to topics like loans or credit cards.

As mentioned above, both Buzzsumo and ahrefs allow you to search for the most shared content by topic and both offer advanced search options.

Further inspiration

There are also several sites I like to look at for inspiration. Whilst these sites don’t give you a great steer on whether or not a particular piece of content was actually successful, with a little digging you can quickly find the original source and pull link and social share data:

  • Visually has a community area where users can upload creative content. You can search by topic to uncover examples.
  • TrendHunter have a searchable archive of creative ideas, they feature products, creative campaigns, marketing campaigns, advertising and more. It’s best to keep your searches broad if you’re looking at this site.
  • Check out Niice (a moodboard app) which also has a searchable archive of handpicked design inspiration.
  • Searching Pinterest can allow you to unearth some interesting bits and pieces as can Google image searches and regular Google searches around particular topics.
  • Reviewing relevant sections of discussion sites like Quora can provide insight into what people are asking about particular topics which may spark a creative idea.

Moving from data to insight

By this point you’ve (hopefully) got a long list of content examples. Whilst this is a great start, effectively what you’ve got here is just data, now you need to convert this to insight.

Remember, we’re trying to answer the questions: “What are websites publishing, and what are people sharing?”

Ordinarily as I go through the creative content research process, I start to see patterns or themes emerge. For example, across a variety of topics areas you’ll see that the most shared content tends to be news. Whilst this is good to know, it’s not necessarily something that’s going to be particularly actionable. You’ll need to dig a little deeper—what else (aside from news) is given coverage? Can you split those things into categories or themes?

This is tough to explain in the abstract, so let me give you an example. We’d identified a set of music sites (e.g. Rolling Stone, NME, CoS, Stereogum, Pitchfork) as target publishers for a client.

Here’s a summary of what I concluded following my research:

The most-shared content on these music publications is news: album launches, new singles, videos of performances etc. As such, if we can work a news hook into whatever we create, this could positively influence our chances of gaining coverage.

Aside from news, the content which gains traction tends to fall into one of the following categories:

Earlier in this post I mentioned that it can be particularly tough to create content which attracts coverage and shares if clients feel strongly that they want to do something directly related to their product or service. The example I gave at the outset was a client who sold insurance and was really keen to create something about insurance. You’re now in a great position to win an argument with data, as thanks to your research you’ll be able to cite several pieces of insurance-related content which have struggled to gain traction. But it’s not all bad news as you’ll also be able to cite other topics which are relevant to the client’s target audience and stand a better chance of gaining coverage and shares.

Avoiding the pitfalls

There are potential pitfalls when it comes to creative content research in that it’s easy to leap to erroneous conclusions. Here’s some things to watch out for:

Make sure you’re identifying outliers…

When seeking out successful pieces of content you need to be certain that what you’re looking at is actually an outlier. For example, the average post on BuzzFeed gets over 30k social shares. As such, that post you found with just 10k shares is not an outlier. It’s done significantly worse than average. It’s therefore not the best post to be holding up as a fabulous example of what to create to get shares.

Don’t get distracted by formats…

Pay more attention to the idea than the format. For example, the folks at Mashable, kindly covered an infographic about Instagram which we created for a client. However, the takeaway here is not that Instagram infographics get coverage on Mashable. Mashable didn’t cover this because we created an infographic. They covered the piece because it told a story in a compelling and unusual way.

You probably shouldn’t create a listicle…

This point is related to the point above. In my experience, unless you’re a publisher with a huge, engaged social following, that listicle of yours is unlikely to gain traction. Listicles on huge publisher sites get shares, listicles on client sites typically don’t. This is doubly important if you’re also seeking coverage, as listicles on clients sites don’t typically get links or coverage on other sites.

How we use the research to inform our ideation process

At Distilled, we typically take a creative brief and complete creative content research and then move into the ideation process. A summary of the research is included within the creative brief, and this, along with a copy of the full creative content research is shared with the team.

The research acts as inspiration and direction and is particularly useful in terms of identifying potential topics to explore but doesn’t mean team members don’t still do further research of their own.

This process by no means acts as a silver bullet, but it definitely helps us come up with ideas.


Thanks for sticking with me to the end!

I’d love to hear more about your creative content research processes and any tips you have for finding inspirational content. Do let me know via the comments.

Image credits: Research, typing, audience, inspiration, kitteh.

Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don’t have time to hunt down but want to read!

Reblogged 4 years ago from tracking.feedpress.it

Spam Score: Moz’s New Metric to Measure Penalization Risk

Posted by randfish

Today, I’m very excited to announce that Moz’s Spam Score, an R&D project we’ve worked on for nearly a year, is finally going live. In this post, you can learn more about how we’re calculating spam score, what it means, and how you can potentially use it in your SEO work.

How does Spam Score work?

Over the last year, our data science team, led by 
Dr. Matt Peters, examined a great number of potential factors that predicted that a site might be penalized or banned by Google. We found strong correlations with 17 unique factors we call “spam flags,” and turned them into a score.

Almost every subdomain in 
Mozscape (our web index) now has a Spam Score attached to it, and this score is viewable inside Open Site Explorer (and soon, the MozBar and other tools). The score is simple; it just records the quantity of spam flags the subdomain triggers. Our correlations showed that no particular flag was more likely than others to mean a domain was penalized/banned in Google, but firing many flags had a very strong correlation (you can see the math below).

Spam Score currently operates only on the subdomain level—we don’t have it for pages or root domains. It’s been my experience and the experience of many other SEOs in the field that a great deal of link spam is tied to the subdomain-level. There are plenty of exceptions—manipulative links can and do live on plenty of high-quality sites—but as we’ve tested, we found that subdomain-level Spam Score was the best solution we could create at web scale. It does a solid job with the most obvious, nastiest spam, and a decent job highlighting risk in other areas, too.

How to access Spam Score

Right now, you can find Spam Score inside 
Open Site Explorer, both in the top metrics (just below domain/page authority) and in its own tab labeled “Spam Analysis.” Spam Score is only available for Pro subscribers right now, though in the future, we may make the score in the metrics section available to everyone (if you’re not a subscriber, you can check it out with a free trial). 

The current Spam Analysis page includes a list of subdomains or pages linking to your site. You can toggle the target to look at all links to a given subdomain on your site, given pages, or the entire root domain. You can further toggle source tier to look at the Spam Score for incoming linking pages or subdomains (but in the case of pages, we’re still showing the Spam Score for the subdomain on which that page is hosted).

You can click on any Spam Score row and see the details about which flags were triggered. We’ll bring you to a page like this:

Back on the original Spam Analysis page, at the very bottom of the rows, you’ll find an option to export a disavow file, which is compatible with Google Webmaster Tools. You can choose to filter the file to contain only those sites with a given spam flag count or higher:

Disavow exports usually take less than 3 hours to finish. We can send you an email when it’s ready, too.

WARNING: Please do not export this file and simply upload it to Google! You can really, really hurt your site’s ranking and there may be no way to recover. Instead, carefully sort through the links therein and make sure you really do want to disavow what’s in there. You can easily remove/edit the file to take out links you feel are not spam. When Moz’s Cyrus Shepard disavowed every link to his own site, it took more than a year for his rankings to return!

We’ve actually made the file not-wholly-ready for upload to Google in order to be sure folks aren’t too cavalier with this particular step. You’ll need to open it up and make some edits (specifically to lines at the top of the file) in order to ready it for Webmaster Tools

In the near future, we hope to have Spam Score in the Mozbar as well, which might look like this: 

Sweet, right? 🙂

Potential use cases for Spam Analysis

This list probably isn’t exhaustive, but these are a few of the ways we’ve been playing around with the data:

  1. Checking for spammy links to your own site: Almost every site has at least a few bad links pointing to it, but it’s been hard to know how much or how many potentially harmful links you might have until now. Run a quick spam analysis and see if there’s enough there to cause concern.
  2. Evaluating potential links: This is a big one where we think Spam Score can be helpful. It’s not going to catch every potentially bad link, and you should certainly still use your brain for evaluation too, but as you’re scanning a list of link opportunities or surfing to various sites, having the ability to see if they fire a lot of flags is a great warning sign.
  3. Link cleanup: Link cleanup projects can be messy, involved, precarious, and massively tedious. Spam Score might not catch everything, but sorting links by it can be hugely helpful in identifying potentially nasty stuff, and filtering out the more probably clean links.
  4. Disavow Files: Again, because Spam Score won’t perfectly catch everything, you will likely need to do some additional work here (especially if the site you’re working on has done some link buying on more generally trustworthy domains), but it can save you a heap of time evaluating and listing the worst and most obvious junk.

Over time, we’re also excited about using Spam Score to help improve the PA and DA calculations (it’s not currently in there), as well as adding it to other tools and data sources. We’d love your feedback and insight about where you’d most want to see Spam Score get involved.

Details about Spam Score’s calculation

This section comes courtesy of Moz’s head of data science, Dr. Matt Peters, who created the metric and deserves (at least in my humble opinion) a big round of applause. – Rand

Definition of “spam”

Before diving into the details of the individual spam flags and their calculation, it’s important to first describe our data gathering process and “spam” definition.

For our purposes, we followed Google’s definition of spam and gathered labels for a large number of sites as follows.

  • First, we randomly selected a large number of subdomains from the Mozscape index stratified by mozRank.
  • Then we crawled the subdomains and threw out any that didn’t return a “200 OK” (redirects, errors, etc).
  • Finally, we collected the top 10 de-personalized, geo-agnostic Google-US search results using the full subdomain name as the keyword and checked whether any of those results matched the original keyword. If they did not, we called the subdomain “spam,” otherwise we called it “ham.”

We performed the most recent data collection in November 2014 (after the Penguin 3.0 update) for about 500,000 subdomains.

Relationship between number of flags and spam

The overall Spam Score is currently an aggregate of 17 different “flags.” You can think of each flag a potential “warning sign” that signals that a site may be spammy. The overall likelihood of spam increases as a site accumulates more and more flags, so that the total number of flags is a strong predictor of spam. Accordingly, the flags are designed to be used together—no single flag, or even a few flags, is cause for concern (and indeed most sites will trigger at least a few flags).

The following table shows the relationship between the number of flags and percent of sites with those flags that we found Google had penalized or banned:

ABOVE: The overall probability of spam vs. the number of spam flags. Data collected in Nov. 2014 for approximately 500K subdomains. The table also highlights the three overall danger levels: low/green (< 10%) moderate/yellow (10-50%) and high/red (>50%)

The overall spam percent averaged across a large number of sites increases in lock step with the number of flags; however there are outliers in every category. For example, there are a small number of sites with very few flags that are tagged as spam by Google and conversely a small number of sites with many flags that are not spam.

Spam flag details

The individual spam flags capture a wide range of spam signals link profiles, anchor text, on page signals and properties of the domain name. At a high level the process to determine the spam flags for each subdomain is:

  • Collect link metrics from Mozscape (mozRank, mozTrust, number of linking domains, etc).
  • Collect anchor text metrics from Mozscape (top anchor text phrases sorted by number of links)
  • Collect the top five pages by Page Authority on the subdomain from Mozscape
  • Crawl the top five pages plus the home page and process to extract on page signals
  • Provide the output for Mozscape to include in the next index release cycle

Since the spam flags are incorporated into in the Mozscape index, fresh data is released with each new index. Right now, we crawl and process the spam flags for each subdomains every two – three months although this may change in the future.

Link flags

The following table lists the link and anchor text related flags with the the odds ratio for each flag. For each flag, we can compute two percents: the percent of sites with that flag that are penalized by Google and the percent of sites with that flag that were not penalized. The odds ratio is the ratio of these percents and gives the increase in likelihood that a site is spam if it has the flag. For example, the first row says that a site with this flag is 12.4 times more likely to be spam than one without the flag.

ABOVE: Description and odds ratio of link and anchor text related spam flags. In addition to a description, it lists the odds ratio for each flag which gives the overall increase in spam likelihood if the flag is present).

Working down the table, the flags are:

  • Low mozTrust to mozRank ratio: Sites with low mozTrust compared to mozRank are likely to be spam.
  • Large site with few links: Large sites with many pages tend to also have many links and large sites without a corresponding large number of links are likely to be spam.
  • Site link diversity is low: If a large percentage of links to a site are from a few domains it is likely to be spam.
  • Ratio of followed to nofollowed subdomains/domains (two separate flags): Sites with a large number of followed links relative to nofollowed are likely to be spam.
  • Small proportion of branded links (anchor text): Organically occurring links tend to contain a disproportionate amount of banded keywords. If a site does not have a lot of branded anchor text, it’s a signal the links are not organic.

On-page flags

Similar to the link flags, the following table lists the on page and domain name related flags:

ABOVE: Description and odds ratio of on page and domain name related spam flags. In addition to a description, it lists the odds ratio for each flag which gives the overall increase in spam likelihood if the flag is present).

  • Thin content: If a site has a relatively small ratio of content to navigation chrome it’s likely to be spam.
  • Site mark-up is abnormally small: Non-spam sites tend to invest in rich user experiences with CSS, Javascript and extensive mark-up. Accordingly, a large ratio of text to mark-up is a spam signal.
  • Large number of external links: A site with a large number of external links may look spammy.
  • Low number of internal links: Real sites tend to link heavily to themselves via internal navigation and a relative lack of internal links is a spam signal.
  • Anchor text-heavy page: Sites with a lot of anchor text are more likely to be spam then those with more content and less links.
  • External links in navigation: Spam sites may hide external links in the sidebar or footer.
  • No contact info: Real sites prominently display their social and other contact information.
  • Low number of pages found: A site with only one or a few pages is more likely to be spam than one with many pages.
  • TLD correlated with spam domains: Certain TLDs are more spammy than others (e.g. pw).
  • Domain name length: A long subdomain name like “bycheapviagra.freeshipping.onlinepharmacy.com” may indicate keyword stuffing.
  • Domain name contains numerals: domain names with numerals may be automatically generated and therefore spam.

If you’d like some more details on the technical aspects of the spam score, check out the 
video of Matt’s 2012 MozCon talk about Algorithmic Spam Detection or the slides (many of the details have evolved, but the overall ideas are the same):

We’d love your feedback

As with all metrics, Spam Score won’t be perfect. We’d love to hear your feedback and ideas for improving the score as well as what you’d like to see from it’s in-product application in the future. Feel free to leave comments on this post, or to email Matt (matt at moz dot com) and me (rand at moz dot com) privately with any suggestions.

Good luck cleaning up and preventing link spam!



Not a Pro Subscriber? No problem!



Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don’t have time to hunt down but want to read!

Reblogged 4 years ago from tracking.feedpress.it

What Deep Learning and Machine Learning Mean For the Future of SEO – Whiteboard Friday

Posted by randfish

Imagine a world where even the high-up Google engineers don’t know what’s in the ranking algorithm. We may be moving in that direction. In today’s Whiteboard Friday, Rand explores and explains the concepts of deep learning and machine learning, drawing us a picture of how they could impact our work as SEOs.

For reference, here’s a still of this week’s whiteboard!

Video transcription

Howdy, Moz fans, and welcome to another edition of Whiteboard Friday. This week we are going to take a peek into Google’s future and look at what it could mean as Google advances their machine learning and deep learning capabilities. I know these sound like big, fancy, important words. They’re not actually that tough of topics to understand. In fact, they’re simplistic enough that even a lot of technology firms like Moz do some level of machine learning. We don’t do anything with deep learning and a lot of neural networks. We might be going that direction.

But I found an article that was published in January, absolutely fascinating and I think really worth reading, and I wanted to extract some of the contents here for Whiteboard Friday because I do think this is tactically and strategically important to understand for SEOs and really important for us to understand so that we can explain to our bosses, our teams, our clients how SEO works and will work in the future.

The article is called “Google Search Will Be Your Next Brain.” It’s by Steve Levy. It’s over on Medium. I do encourage you to read it. It’s a relatively lengthy read, but just a fascinating one if you’re interested in search. It starts with a profile of Geoff Hinton, who was a professor in Canada and worked on neural networks for a long time and then came over to Google and is now a distinguished engineer there. As the article says, a quote from the article: “He is versed in the black art of organizing several layers of artificial neurons so that the entire system, the system of neurons, could be trained or even train itself to divine coherence from random inputs.”

This sounds complex, but basically what we’re saying is we’re trying to get machines to come up with outcomes on their own rather than us having to tell them all the inputs to consider and how to process those incomes and the outcome to spit out. So this is essentially machine learning. Google has used this, for example, to figure out when you give it a bunch of photos and it can say, “Oh, this is a landscape photo. Oh, this is an outdoor photo. Oh, this is a photo of a person.” Have you ever had that creepy experience where you upload a photo to Facebook or to Google+ and they say, “Is this your friend so and so?” And you’re like, “God, that’s a terrible shot of my friend. You can barely see most of his face, and he’s wearing glasses which he usually never wears. How in the world could Google+ or Facebook figure out that this is this person?”

That’s what they use, these neural networks, these deep machine learning processes for. So I’ll give you a simple example. Here at MOZ, we do machine learning very simplistically for page authority and domain authority. We take all the inputs — numbers of links, number of linking root domains, every single metric that you could get from MOZ on the page level, on the sub-domain level, on the root-domain level, all these metrics — and then we combine them together and we say, “Hey machine, we want you to build us the algorithm that best correlates with how Google ranks pages, and here’s a bunch of pages that Google has ranked.” I think we use a base set of 10,000, and we do it about quarterly or every 6 months, feed that back into the system and the system pumps out the little algorithm that says, “Here you go. This will give you the best correlating metric with how Google ranks pages.” That’s how you get page authority domain authority.

Cool, really useful, helpful for us to say like, “Okay, this page is probably considered a little more important than this page by Google, and this one a lot more important.” Very cool. But it’s not a particularly advanced system. The more advanced system is to have these kinds of neural nets in layers. So you have a set of networks, and these neural networks, by the way, they’re designed to replicate nodes in the human brain, which is in my opinion a little creepy, but don’t worry. The article does talk about how there’s a board of scientists who make sure Terminator 2 doesn’t happen, or Terminator 1 for that matter. Apparently, no one’s stopping Terminator 4 from happening? That’s the new one that’s coming out.

So one layer of the neural net will identify features. Another layer of the neural net might classify the types of features that are coming in. Imagine this for search results. Search results are coming in, and Google’s looking at the features of all the websites and web pages, your websites and pages, to try and consider like, “What are the elements I could pull out from there?”

Well, there’s the link data about it, and there are things that happen on the page. There are user interactions and all sorts of stuff. Then we’re going to classify types of pages, types of searches, and then we’re going to extract the features or metrics that predict the desired result, that a user gets a search result they really like. We have an algorithm that can consistently produce those, and then neural networks are hopefully designed — that’s what Geoff Hinton has been working on — to train themselves to get better. So it’s not like with PA and DA, our data scientist Matt Peters and his team looking at it and going, “I bet we could make this better by doing this.”

This is standing back and the guys at Google just going, “All right machine, you learn.” They figure it out. It’s kind of creepy, right?

In the original system, you needed those people, these individuals here to feed the inputs, to say like, “This is what you can consider, system, and the features that we want you to extract from it.”

Then unsupervised learning, which is kind of this next step, the system figures it out. So this takes us to some interesting places. Imagine the Google algorithm, circa 2005. You had basically a bunch of things in here. Maybe you’d have anchor text, PageRank and you’d have some measure of authority on a domain level. Maybe there are people who are tossing new stuff in there like, “Hey algorithm, let’s consider the location of the searcher. Hey algorithm, let’s consider some user and usage data.” They’re tossing new things into the bucket that the algorithm might consider, and then they’re measuring it, seeing if it improves.

But you get to the algorithm today, and gosh there are going to be a lot of things in there that are driven by machine learning, if not deep learning yet. So there are derivatives of all of these metrics. There are conglomerations of them. There are extracted pieces like, “Hey, we only ant to look and measure anchor text on these types of results when we also see that the anchor text matches up to the search queries that have previously been performed by people who also search for this.” What does that even mean? But that’s what the algorithm is designed to do. The machine learning system figures out things that humans would never extract, metrics that we would never even create from the inputs that they can see.

Then, over time, the idea is that in the future even the inputs aren’t given by human beings. The machine is getting to figure this stuff out itself. That’s weird. That means that if you were to ask a Google engineer in a world where deep learning controls the ranking algorithm, if you were to ask the people who designed the ranking system, “Hey, does it matter if I get more links,” they might be like, “Well, maybe.” But they don’t know, because they don’t know what’s in this algorithm. Only the machine knows, and the machine can’t even really explain it. You could go take a snapshot and look at it, but (a) it’s constantly evolving, and (b) a lot of these metrics are going to be weird conglomerations and derivatives of a bunch of metrics mashed together and torn apart and considered only when certain criteria are fulfilled. Yikes.

So what does that mean for SEOs. Like what do we have to care about from all of these systems and this evolution and this move towards deep learning, which by the way that’s what Jeff Dean, who is, I think, a senior fellow over at Google, he’s the dude that everyone mocks for being the world’s smartest computer scientist over there, and Jeff Dean has basically said, “Hey, we want to put this into search. It’s not there yet, but we want to take these models, these things that Hinton has built, and we want to put them into search.” That for SEOs in the future is going to mean much less distinct universal ranking inputs, ranking factors. We won’t really have ranking factors in the way that we know them today. It won’t be like, “Well, they have more anchor text and so they rank higher.” That might be something we’d still look at and we’d say, “Hey, they have this anchor text. Maybe that’s correlated with what the machine is finding, the system is finding to be useful, and that’s still something I want to care about to a certain extent.”

But we’re going to have to consider those things a lot more seriously. We’re going to have to take another look at them and decide and determine whether the things that we thought were ranking factors still are when the neural network system takes over. It also is going to mean something that I think many, many SEOs have been predicting for a long time and have been working towards, which is more success for websites that satisfy searchers. If the output is successful searches, and that’ s what the system is looking for, and that’s what it’s trying to correlate all its metrics to, if you produce something that means more successful searches for Google searchers when they get to your site, and you ranking in the top means Google searchers are happier, well you know what? The algorithm will catch up to you. That’s kind of a nice thing. It does mean a lot less info from Google about how they rank results.

So today you might hear from someone at Google, “Well, page speed is a very small ranking factor.” In the future they might be, “Well, page speed is like all ranking factors, totally unknown to us.” Because the machine might say, “Well yeah, page speed as a distinct metric, one that a Google engineer could actually look at, looks very small.” But derivatives of things that are connected to page speed may be huge inputs. Maybe page speed is something, that across all of these, is very well connected with happier searchers and successful search results. Weird things that we never thought of before might be connected with them as the machine learning system tries to build all those correlations, and that means potentially many more inputs into the ranking algorithm, things that we would never consider today, things we might consider wholly illogical, like, “What servers do you run on?” Well, that seems ridiculous. Why would Google ever grade you on that?

If human beings are putting factors into the algorithm, they never would. But the neural network doesn’t care. It doesn’t care. It’s a honey badger. It doesn’t care what inputs it collects. It only cares about successful searches, and so if it turns out that Ubuntu is poorly correlated with successful search results, too bad.

This world is not here yet today, but certainly there are elements of it. Google has talked about how Panda and Penguin are based off of machine learning systems like this. I think, given what Geoff Hinton and Jeff Dean are working on at Google, it sounds like this will be making its way more seriously into search and therefore it’s something that we’re really going to have to consider as search marketers.

All right everyone, I hope you’ll join me again next week for another edition of Whiteboard Friday. Take care.

Video transcription by Speechpad.com

Sign up for The Moz Top 10, a semimonthly mailer updating you on the top ten hottest pieces of SEO news, tips, and rad links uncovered by the Moz team. Think of it as your exclusive digest of stuff you don’t have time to hunt down but want to read!

Reblogged 4 years ago from tracking.feedpress.it