How do you slice data to find what links are worthy?
Many SEOs just pull up OSE, and sort from highest PA. In the past, this was a fairly reliable way to find the strongest links first. But OSE’s data is pretty useless at this point, we have practically eliminated it from our analysis.
The only thing we care about is finding the highest PR contextual links from domains that aren’t penalized.
“Metrics”
Whether it’s PA, DA, Cemper Power, Cemper Trust, Cemper PowerTrust, etc. it is important to remember they are all just a “best guess” and none reflect when site has a penalty. Even PR doesn’t reflect penalties.
Nerdy breakdown:
The only difference in the metrics is their level of “exponential value” meaning how they scale the data. A simplistic example would be 2^3=8 on one URL, versus 2^5=32 on another URL. So it may be that with all things equal, 3 links generate a score of 8, and 5 links generate a score of 32. One company’s interpretation of the data says “2 more links means it is 4x more powerful” (32/8=4) but we know that’s not true. Another company might interpret the data as “compared to the rest of the link graph of the web, these are within 24%-40% power of each other. When scaled to the big picture they are only 2% apart.”
But what can you do with that? Practically nothing. This is why we ignore so many of the metrics and only focus on the ones we can verify, which is “does it rank?”
Sidenote: The only new “cool” metric with potential is Cemper Trust. It is based on the idea of PageTrust, meaning “hand-picked trusted verified sources have a score of 100 or PR10.” It’s basically a manual redistribution of juice based on the top 100 domains. It’s a way of saying “how far away are you from something we know is trustworthy?” So far in our testing, this metric seems to just show correlation and no causation. Watch it, but don’t act because of it.
Link Research Tools
It’s expensive. It’s worth it. Unfortunately the day pass we loved to suggest to members is now gone. If members want to buy an account and share it to keep costs down, that is an avenue you could take. Or maybe you can pay someone for a “link analysis” where they will just run reports for you and give you the CSV files. We are also considering a few new avenues to provide this data to our membership. APIs for LinkResearchTools are very expensive and would drive up the cost of the membership.
Account Type Differences
We are on the Superhero plan and we still can’t burn up all of our credits each month, even with monthly recurring tasks and bookmarklets for quick checks. For $500/month we think it can safely handle 5 normal SEOs.
The starter plan would be too much of a bottleneck for most users. The “backlink profiling” number is the amount of times you can use the BLP tool, which is the Lamborghini of link analysis. The starter plan only allows for 10 BLP checks for $130 ($13/check). Compare that to sharing a Superhero plan, 250 BLP checks for $500 ($2/check). Divided up among five people would it would be 50 checks each for $100. We are quite positive LinkResearchTools doesn’t want us recommending you to share an account with others (or paying others for them running reports) but it is simply too expensive for some members’ budgets and we are trying to help them out. Being that the day pass is gone, we don’t have anything less than $100 to recommend.
Certifications
Link Research Tools has certifications for knowledge of their tools. For the associate level, it is an eight hour webinar series that will show you how all of the tools function. The test is timed, so you have to know exactly which tools to run based on the data the questions are looking for.
It is important to remember Cemper’s analysis (as brilliant as he is) during the webinar series is not congruent with our philosophies. The way he slices data to “good links” and “bad links” is not ideal for SEO. He is heavily promoting his Link Detox tool, which in theory is very powerful, but in practice doesn’t help anything rank.
The certification is worth it, just so you know how all of the tools work and function together. It saved us at least eight hours each month.
An Example:
One of the questions on the test asked which link provided the most juice and which link provided the least amount of juice to a particular URL. There were only 3 links to choose from.
The toolset it required you to use to get to this point only showed Cemper’s PowerTrust metrics, which is what the question wanted you to answer based on their data.
Our response was basically ripping a new one for teaching bad SEO. The site was a .com trying to rank in the US. One link was from Wikipedia, one from another .com, and one from a .co.uk. The answer Cemper was looking for was:
“Wikipedia is the strongest, and the .com is the weakest. Try to get a link in Wikipedia and ignore the .com.”
In reality, that is horrible prioritization for acquiring links. Wikipedia is nofollow, but the tool didn’t show that. The .co.uk isn’t going to push a .com that much in the US (and it may have been a blogroll link). The .com was actually the only one remaining by default worth going after. So really, our answer is the exact opposite of what they wanted. We still passed with flying colors because we explained our reasoning.
BLP
This is the only tool you need out of the entire toolset. Always check every box on the “start report” screen. It doesn’t save you any credits, just time. It only takes a few minutes to load even with every checkbox selected. Skip sitewide links after 5, and don’t bother adding 2x, 3x, etc. to it. It only eats up credits. Be sure to analyze the entire domain (not just a page or URL) to get a clear picture of what is happening to the site.
Once your report is finished, experiment with the filters. We like to show only followed links and mentions, with the “To URL” field shown. In low competition niches we start with minimum PR0 (to eliminate PR N/A links) and sort descending according to PR.
From this point, usually only 20% of the links are remaining. If it’s a medium-competition niche, we bump up the minimum to PR1 and it comes down to about 10% of links. It’s usually safe to filter out any footers or sidebar links at this point too, and get down to the 5% of links that really push the site.
Using the Data
From here you can sort the small amount of data remaining a number of ways. The first is to get an overall count of how many links you need to beat them, and how strong those links need to be. If the competitor has 50 links remaining, and 40 of those are PR1, you are in a great position. You can get 10 PR3+ links and usually outrank them no problem. It doesn’t matter if they have 50,000 links pointing to their site, the right 10 can beat it. Just set your goal.
Steal links where you can. Sometimes it’s a professional directory that hasn’t been nailed by Panda. Sometimes it’s just an article referencing informational resources in your niche, and your competitor and all the other competitors are listed. Grab all of those links. They are just low-hanging fruit. If you don’t grab these links, you are lazy.
Ahrefs
Our second choice for link analysis is Ahrefs. It is the only tool that really gives a beautiful graph of how hard someone is promoting their site over time. Their data is updated constantly. The data for new and lost links is awesome. And best of all there is a free plan which is great if you are trying to save on monthly expenses.
Their metric for “which links have the most power” is pretty solid. It’s not accurate as discussed earlier, but it is one of the better ones.
What if I can’t afford Link Research Tools, can I do it by hand?
Absolutely. It takes forever and the data set will be smaller.
1. Get a link list from somewhere like OSE or Ahrefs. Try to get it in a .CSV file.
2. Hopefully that list has a column of nofollow links.
3. Learn Excel/Numbers filtering systems to kill all nofollow links.
4. Run a mass PR checking tool on the remaining links. There are plenty of free ones online.
5. Add that data to your set.
6. Filter it down some more, maybe to only PR1.
7. Get to work.
It’s going to take a very long time. Your data is probably going to get errors in it. But if you are in a low-competition market, you can get enough data to for real action steps.
Analysis to Action
For every 15 minutes of analysis, you need to be able to take hours of action. In 15 minutes, you should be able to:
1. Set a goal of how many links you need to build to beat the top competitor
2. Have some links you can easily steal to easily work towards that goal
3. Set a budget and timeframe for beating the top competitor
There is no point in spending hours analyzing the data. It won’t make you rank any higher. Just get to work on acquiring the strong links.
Spend more time analyzing the top competitor and less time analyzing your own site. Once you have nailed the top competitor, you might want to attack the second or third competitor over the next few months just to secure your top spots. Or if you get to number one, work on your second and third sites. But the bottom line is no amount of overanalyzing will help you rank.