TL;DR We’ve re-ranked the Inc. 500 for the past two years based on each site’s propensity to see a loss or gain in Organic Search visibility. It’s a thought experiment that has yielded a model, allowing us to use a combination of domain-level link metrics to predict a site’s likelihood to succeed or fail in Natural Search. We’ve published a detailed study on our findings called the Vector Report.
Admittedly, I am a data junkie. I use, or have used, nearly every SEO tool there is. Having access to so many tools and metrics, it’s easy to get analysis paralysis. Should I use Moz for this project? Or should I use Ahrefs? Perhaps I should consider Majestic, Searchmetrics or CognitiveSEO?
Naturally, all of those tools have their strong suits, but they all also come with their own specific datasets as each link index represents the chosen pocket of the web that the provider crawls. Moz is, of course, the pioneer in our field and people talk more about Domain Authority than any other available metric. However, all of the available tools have their strengths, but which metrics or combination of metrics yields the most accurate predictor of performance?
We Wanted to determine which Link Metrics are the Best Predictors of Visibility
In efforts to get to the bottom of that question, we gathered all of the available domain-level link data and compared it against data in Searchmetrics’ fantastic database. For those that don’t know, the Searchmetrics platform collects and exposes a visibility measure for millions of domains. On a weekly basis, they also publish lists of the top 100 losers and top 100 winners in the visibility race across their database.
So we used their Winners and Losers lists as a corpus to predict against. We collected every available domain level metric from Moz, Majestic, Ahrefs, SEMRush, CognitiveSEO and Searchmetrics for the domains in the Searchmetrics Winners and Losers lists. For those keeping score at home, there are 116 of them. Then we thought through what those metrics represent and ended up omitting variables that didn’t quite make sense. For instance, it doesn’t make sense to look at the Searchmetrics visibility score because it is a variable that is a result of visibility rather than a cause for visibility.
Once we settled on what could be considered valuable metrics, we used a series of statistical models until we identified the series of metrics that best correlated with the results in Searchmetrics’ Winner and Losers lists.
Determining the model itself took a lot of effort from our Research Analyst (aka Data Scientist) Vicky Qian (aka Vicky Money). A sample size of 200 sites is not large, so the model required a lot of refinement and ultimately she ended up using 5-fold cross validation to reduce the variance after using random forest and lasso to identify the series of variables to use in her final logistic regression.
After we developed a solid model, we ran that model on all 500 sites in the Inc. 500 for both 2014 and 2015 and that’s how we arrived at our rankings. As a final check to ensure the validity of the model, we also reviewed the visibility trend of these domains in Searchmetrics as well.
As you can see in the plot, our model predicted an overwhelming amount of the sites as losers in that their probability of being visible in Organic Search is less than 0.5. This is because the sites that in our training dataset of winners and losers differ from sites in the Inc. 500. Deeper analysis uncovered that much of the Inc. 500 isn’t very active in Organic Search and, generally speaking, the data indicates that you can expect these more established sites will typically outperform sites that appear on the Inc. 500.
So rather than taking these values absolutely in our scoring model, we’ve ranked the sites relatively within the Inc. 500.
The Winning Combination of Link Metrics
The combination of link metrics that we’ve found ultimately best predicted visibility are:
- Moz’s MozRank
- Ahrefs Referring C Class’s
- Majestic Citation Flow
- Majestic Topical Trust Flow
- SEMRush Rank
- CognitiveSEO Unnatural Links
- CognitiveSEO OK Links
- Moz Spam Score
It’s great to see Moz’s Spam Score, Majestic’s Topical Trust Flow and Cognitive’s Link Classfiier metrics show up in our most accurate model. Personally, I find it harder to trust newer metrics that haven’t been quite battle-tested yet so this outcome helps reinforce their value.
We Wanted to Empower Inc. 500 Companies
The Inc. 500 a subset of is Inc Magazine’s list of the 5000 fastest growing private companies. While it’s the Fortune500 that sounds impressive on a slide in a presentation or in a casual conversation, it’s often the Inc. 500 companies that are trying to push hard to innovate.
In addition to our Fortune500 clients, we love working with these types of clients and figured that this analysis could potentially help out a lot of them at once. It also wouldn’t hurt if our research started some conversations between our organizations.
The data indicates that many of these companies may not have SEO top of mind so we wanted to put together something that helps inform how they should be approaching SEO to be more effective. Who knows, maybe bringing Organic Search into their marketing mix might get them higher on the list.
We’ve performed our analysis twice for two reasons. One, to check the validity of our model year over year and two, because this type of work takes a long time as a side project! That said, rankings for both the 2014 and 2015 Inc. 500 are available. You can find the full results in our rather hefty whitepaper. However, in the sneak preview below, you can see that the 2015 Inc. 500 set is outperforming 2014 for the key segments of link metrics that we’ve reviewed.
Generally speaking, Inc. 500 companies suffer from link popularity and trust deficiencies across the board. Year over year, the Inc.500 has turned over pretty dramatically and more unnatural links sprouted up than in 2014. Nevertheless, within our grading system, we saw a 38% increase in companies receiving higher grades year over year.
Additionally, we go into details on findings for ten of the verticals and highlight two websites per grade range to help you better understanding how to capitalize on optimization opportunities.
Now to give you a sense of what we looked at.
Why Visual.ly is #1 (Inc. 500 #174)
Recently acquired Visual.ly is a content creation platform for data visualization. They boast the ability to deliver premium eBooks, presentations, interactives and videos with their marketplace of artists. However, what many marketers may recognize them as an infographic repository that allows users to showcase their work. Visually was acquired earlier this this year by ScribbleLive for an undisclosed amount.
Now, that’s all well and good, but why are they the outlying dot above 0.7 in our data plot above? The answer is simple, yet it’s still the answer that no one wants to hear – Content.
Frankly, infographics are one of the most powerful pieces of content for link acquisition and Visual.ly is literally a site full of them. Also, giving users public portfolios is a good means of turning your user base into a link building army for you; #3 rated site, Contently is another good example of this tactic at play. However, sites should be careful of this tactic as it appears that many users will build spammy links to their portfolio page to make the page more powerful for linking back to their other sites. Look no further than one of the most linked to pages on the site for an example of that.
In reviewing the site in CognitiveSEO’s Link Visualization tool, you can see that there is a wealth of links to a variety of link targets throughout the site rather than just to the homepage and a small number of internal pages. This distribution of links is healthy to the flow of link equity and is an indication that many pages across the site have authority.
Also, the visualization highlights both Unnatural and Suspect links, as determined by CognitiveSEO’s Unnatural Links classifier, by coloring them in red and orange respectively. There are very few specks of orange here and no specks of red. That’s an indication that the link profile is quite healthy.
In reviewing the link velocity, you see that Visual.ly has continued to sustain their growth in links over time. Throughout that time, the number of links has mostly stayed in lockstep with the growth of linking root domains aside from late 2014. That is to say that Visual.ly has maintained a diverse link profile.
Going deeper into the Unnatural Links classification, only 1% of the sample of links reviewed have been pegged as Suspect. With the percentage being so small, it’s likely that any potential spam currently in the profile will be largely outweighed by the good links.
The site’s anchor text is 43% percent branded with the other 57% being mostly miscellaneous anchors. There is some commercial anchor text in there, such as “infographics software” and “data visualization software,” but in the grand scheme of the links to the site, they aren’t enough to likely trip a spam filter.
Perhaps more than any other site on the list, the distribution of trust across from the domains linking to the Visual.ly is quite strong and natural with more than enough at the very top end to outweigh any of the potential spam previously discussed.
Visually is the highest rated site on the list because their product focus is on exposing content that numerous audiences find valuable and by building a user base motivated to get eyeballs on their work.
Why ServeSys is Dead Last (Inc. 500 #462)
On the opposite end of the spectrum, ServeSys is not doing so hot. In fact, I just gave them that link to help dilute the issues in their link profile. Truthfully though, who am I to judge? They made $4.7 million in revenue in 2014 as an IT infrastructure and managed services organization, so clearly they know what they are doing. However, from an SEO and, more importantly, Content Strategy perspective they could use some love.
Before someone with Marketing in their title can say, “What are you talking about? We installed the Yoast SEO plugin,” the ServeSys site, like many other on the Inc. 500 is a brochure. My best guess is that this site is meant to be what prospects find on Google after they receive a cold call to prove that it’s a real company. That’s because there’s not much in the way of content that would support a lead generation effort. Also, there’s no indication that an analytics package is installed either; so if there is a lead gen effort happening, it’s not being measured.
If these things weren’t enough to indicate that there hasn’t been much of an digital marketing effort, at least with this property, the SEO components are a bit more revealing.
As I was typing, I thought that I’d forgotten to paste the visualization of the link profile below. Turns out I hadn’t, there’s just not many links or link targets. Turn your head to the left about 30 degrees if you can’t see it.
Of that handful of links, one of them has been pegged by the CognitiveSEO link classifier as Unnatural.
The site has also struggled to generate a consistent link velocity. Given the lack of editorial content, it’s curious that the site was able to generate inbound links at all.
The anchor text to the site is 92% branded which would be absolutely fantastic if the volume was there. However in this case, we’re talking about seven links.
The small amount of links in the profile makes it especially bad that one of them is an unnatural web directory link. It’s not so much a question of whether this site will get penalized, rather it’s an issue of ServeSys.net never having visibility in the first place.
Finally, the trust in the link profile is heavily on the low end and the volume is not competitive. Visibility is wholly not in the cards for ServeSys.net and that’s why they carry the worst grade on the Inc. 500.
I suspect that, for ServeSys, digital marketing is not a huge concern. It’s very likely that they get tons of referral business and they have an outbound sales team that makes it happen for them. However, it seems like a disconnect for an IT firm to not be more heavily invested in their website.
If they do want to get on the board, I’d recommend building out their site a bit more to include case studies, educational and thought leadership content. In fact, ServeSys might want to take a page from Visual.ly’s book and consider infographics and eBooks, if it makes sense for their audience. Also, since they likely have a number of relationships with IT vendors in the space, ServeSys should consider writing testimonials for those brands because many of those companies have high authority sites and will include a link back. Finally, if the site is meant to support lead generation, it should feature a stronger call to action for users to contact and also get rid of the cool, but distraction ParticlesJS visualization.
See the Rankings
We’ve exposed the rankings for all the review at our Vector Report landing page. Over there you can see where these companies rank on the 2015 Inc. 500 and what their ranking is based on our model from an SEO perspective.
Get Your Company’s Custom Report
If you’re on the illustrious Inc. 500, we’ve already prepared a customized report for you, about your site, based on our findings. In fact, you’ll probably receive it from us soon if you haven’t already. Or you could request your report. If you’re not on the Inc. 500, but are curious as to where you’d rank based on your link profile, feel free to reach out and we can whip up a custom report for you as well.
Download the Whitepaper
There’s learnings for everyone in our 144-page whitepaper. We encourage you to download it and thumb through it until you find something that you can use to improve your own SEO. We specifically focus on what content types work best for the verticals that we reviewed. So you might find some insights to inform Content Strategy for the remainder of the year and going into 2017.
Special Thanks to CognitiveSEO
I’ve long been a fan of CognitiveSEO, due to their agility and focus on innovation. When I came up with the idea to do this study, I knew it’d make a valuable addition. The good folks at CognitiveSEO were very helpful in granting us reasonably unfettered access to their system twice and I’d encourage you to check out a free trial of the tool right now if you haven’t. It’s a fantastic (and as we learned, highly accurate) way to determine and visualize the health of your link profile.
There’s More to Come
As you might imagine, we’ve got a wealth of insights from reviewing these 500 sites. Keep your eye on our blog for more of our findings and strategies for overcoming the obstacles plaguing these sites.
But for now, that’s all I’ve got. See you in the comments!