In SEO, instinct is impor­tant, but data rules all. What fol­lows is a method to help focus brain­storm­ing, pro­vide insight to com­peti­tor strate­gies, and help you know where to focus your con­tent cre­ation ener­gies when the goal is links. You’ll still need to use your instincts and intu­ition for top­ics that haven’t already been cov­ered, but even then you’ll be doing it with a keen­er knowl­edge of what types of con­tent peo­ple are link­ing to in your indus­try.

What we are going to look at in this arti­cle is a way to use link data to deter­mine what con­tent res­onates best among those who link to help dri­ve con­tent ideas to serve as link bait. The tool I’ll use to get this link data is Linkdex, which spon­sors Momen­tol­ogy and was recent­ly named Best SEO Soft­ware at the 2015 U.S. Search Awards. The Linkdex plat­form does a lot of things extreme­ly well, includ­ing aid­ing in link build­ing. It doesn’t tell you what to do, but pro­vides a robust array of well-orga­nized data to help you make your own deci­sions. To cre­ate this guide I’ve used sites that don’t com­pete with any of my clients. My selec­tion was based on sites rank­ing for the phrase “why the heck is kim kar­dashi­an famous”. The “client” will be Cele­buzz, and its com­peti­tor is TMZ. Assum­ing I’m work­ing with Cele­buzz (again, I’m not), let’s look at how we can use our link data to iso­late what Cele­buzz con­tent is attract­ing links.

Step 1: Get Your Data

The first step of the process is col­lect­ing your data. To do this we first view our back­link data: From here we need to export that data so we have it in a spread­sheet. To do this we sim­ply need to click the Export but­ton: When you’re giv­en the options you can select them based on how you want to con­di­tion your data. If you select one page per domain you will miss data where the same domain links to mul­ti­ple pages; how­ev­er if you select all pages (as I will be doing here) you’ll get this data how­ev­er if there are run-of-site links or oth­er issues like blog tags you will get inflat­ed num­bers report­ing for a page. We will skew in that direc­tion here. After down­load­ing the data, you will have a spread­sheet with buck­ets of data in it. A lot of this can be inter­est­ing but for our pur­pos­es here we can clear all but two columns: the source URL and tar­get URL. As we dis­cussed above, we may end up with issues from tag and cat­e­go­ry pages. To deal with this you can sim­ply fil­ter the spreadsheet’s col­umn A (source URL) to pages that have the word “tag” or “cat­e­go­ry” in them and remove those that are indeed dupli­cates in this way. This will leave us with a spread­sheet sim­i­lar to:

Step 2: Making Your Data Useful

Sort­ing through the 67,617 rows of data look­ing for pat­terns might be a bit time con­sum­ing. What we want to dis­cov­er is what sub­jects work best as a whole. Let’s get our data into a for­mat that tells us this. Now let’s move the tar­get URL col­umn into a Notepad doc­u­ment. It will look some­thing like: We will then do a find and replace. You’ll look through your URLs to find those that are appro­pri­ate to you (based on the struc­ture and what you want includ­ed). For exam­ple, here the words are sep­a­rat­ed by dash­es, in some it may be under­scores or oth­er char­ac­ters. In our exam­ple we will do a find-and-replace for the URL. Because Cele­buzz uses sub­do­mains (such as kimkardashian.celebuzz.com) we will have to decide whether we want these includ­ed. I tend to remove them, but there are times they can be use­ful. We’ll also want to sep­a­rate the words out of the page names by replac­ing the dash­es and slash­es with spaces. This will give us: The next step is to save this doc­u­ment as an HTML file and open it in either Fire­fox or Chrome because you’re going to need the SEO Quake plu­g­in for the next step. This will leave you with a page that looks like:

Step 3: Finding What Types Of Content Work

The final step in the process is find­ing what works out of this mess of words. Click the SEO Quake key­word den­si­ty but­ton: With this we will be pre­sent­ed with a list of phras­es ordered by the num­ber of times they appear. First we’ll see the sin­gle words, then the two-word sets, and then the three- and four-word sets. We’re also giv­en a key­word cloud. The page essen­tial­ly looks like: This is the data we real­ly want. What we can now see is a break­down of the most used words and that tells us a lot about what con­tent is res­onat­ing and what types of con­tent attract links. Whether the point of inter­est is that “pho­tos” attract a large num­ber of links, that pieces from 2012 have more links than any oth­er year or that the word “Kar­dashi­an” is the name most used of all the peo­ple with Kim Kar­dashi­an con­tent attract­ing almost 3x the num­ber of links as Justin Bieber we can start to make some sol­id deci­sions regard­ing what type of con­tent to devel­op if links are our goal. Worth not­ing as well, if you want­ed to restrict this to pages from a set time peri­od (for exam­ple) in the past year or if your busi­ness is sea­son­al, select­ing a time­frame like November/December you can sort the spread­sheet by URL or fil­ter it to only include the time ranges you are inter­est­ed in. This works only on sites that have date-spe­cif­ic por­tions in their URLs. As not­ed above, this doesn’t just work for your own con­tent but in see­ing what’s work­ing for oth­ers. While I won’t force you through all the steps and screen­shots again, the same process done on TMZ pro­duces: Here we’re see­ing that pho­tos are also high on the list, as is Kar­dashi­an, but there appears to be a broad­er dis­tri­b­u­tion of terms. Mov­ing to the two-key­word phras­es we can see that Justin Bieber and Lamar Odon are both linked to more than Kim Kar­dashi­an. This illus­trates some new oppor­tu­ni­ties and look­ing into the types of arti­cles on Justin and Lamar that are attract­ing the links and how they’re push­ing them out will help dri­ve a con­tent-based link build­ing strat­e­gy based not on guess­ing but on what’s work­ing.

Not The End All Be All

All of this is based on pre-exist­ing data that can pow­er some SEO wins. You don’t have to stretch your imag­i­na­tion too far to know that new sub­jects come up in vir­tu­al­ly every field and that this tech­nique wouldn’t tell you it was a good idea until it’s too late. In those cas­es, you’ll still need to rely on your instinct when cre­at­ing your con­tent.