I was recently inspired by a post by MikeTek, Examining the Top 150 In-Linked Posts at SEOmoz. While the information was very informative, it was taken from a limited sample of the over 3,800 post on SEOmoz. Along with the small sample size, the data was not proportional to other categories and posts. Mike encouraged anyone with a little automation skills to look into all the data on SEOmoz and really look into what makes a post link worthy.
For this post we will be looking only at In Linking Domains (ILDs) which is a great way to determine if a post is popular throughout the net. For example if you have a site like dummy-domain.com that links to your post 1,000 times throughout its site, it would count as 1 ILD. A great viral post will have a large number of ILDs along with a large number of links.
LET'S START WITH THE DATA
Much like Mike’s data, there were some categories that collected more links than others. In fact there were five categories that have the most domains linking to them, Link Building (1,448 domains), Google (1,419 domains), Technical Issues (1,243 domains), Miscellaneous (1,215 domains), and Whiteboard Friday (1,044 domains). See the graph below for the top 30 of the SEOmoz categories.
The data in the chart above can be a little misleading because some of those categories have many more posts than others do. To make all the data proportional, I took the total number of linking domains and divided it by the number of posts. The chart changes considerably when the data becomes normalized. From the chart below you can see that the top 3 categories are Webdev (19), Technical Issues (14), and On-Page Issues (10).
From the charts above it’s easy to say that “Link Building” is a very popular blog topic but it doesn’t always draw in the links like other topic can. Part of that could be because of the difference in options on link building techniques. Topics like “Technical Issues” and “On-Page Issues” the types of content that most people will agree on and possibly want to share with others in the business.
To take the study a little further, I stored all the post’s title’s and created a title word cloud for the top 10% of the 3,800. This hopefully will give you an idea of what topics could have the possibility of being link worthy in your future post. I was going to try and come up with a “Super” title based on the words in the cloud but couldn’t come up with anything catchy. Maybe one of you Mozzers can come up with something amazing from the cloud.
Like Mike did in his previous post, let’s take a look at the content in the posts: images, list, and videos. In the chart below you will see that having a just a list in a post compared to just text, doubles the average number of linking domains. Have a video compared to just text will almost triple the average number of linking domains. A post that has an image and a list will also triple the average number of linking domains to the post.
I’m sure many of you are like me and do a quick scan of the post before actually reading it. By adding images, videos, and lists, it makes it easy to get a quick synopsis on what the post is about, encouraging people to go back and do a full read along with a possible link. Adding Images and Lists are easy to do and could result in a post that is more link worthy.
I also recorded the length of the post to see if it had an effect on the average number of linking domains. The length recorded was only that of the post and not the comments or other areas of the page to keep the data accurate. I’ve read that most blog post should be kept to 500 words or less. That information seems to be incorrect if you are going to post on SEOmoz and want it to be link worthy. The chart below shows that posts with 1800 or more words have a much higher average of linking domains.
BIGGEST TAKEAWAYS
I feel the real take away from this post is in the last two graphics summarized below:
- Content is most important thing to a posts but posts with extra visual content attract extra links.
- By adding simple visual content, like lists and images, can increase the number of ILDs by good percent.
- Posts with videos included will attract almost 3 times more ILDs than a plain text post.
- Posts with all three media types (videos, images, and lists) will attract almost 6 times more ILDs than a plain text post.
- Contrary to common beliefs, large posts seem to attract more links than posts with 900 words or less.
- Posts with between 1800 and 3000 words will attract more than 15 times more ILDs than a post with less than 600 words.
The first part of this study was only with SEOmoz data and in the next part of this study I have decided to take on a huge project and taking a look at some of the top SEO/Internet Marketing blogs on the web. With a larger sample size we may be able to find out if the information found during this study will hold true in other areas. Stay tuned!
Excellent work!
Not sure about the part that videos help people get a quick synopsis of a post though. If anything I'd argue that videos slow people down. What takes 60 seconds to say can be skimmed in 5. Maybe that's a good thing?
You are correct, a video might not give you a quick synopsis of the post but I often start the video and open another tap to do work, if I hear something interesting then I filp back to see what was going on. I'm lazy and sometimes don't like to read so a video is a nice treat on my brain. =)
yeah me to, unless my "super fast" broadband connection isn't being particulaly super fast...
I'm a reader vs. a viewer myself. I'm sure I'm in the minority, but I tend to skip videos if I don't have the time then and there, vs. with written copy, I leave it up in a tab in my browser until I get to it.
From what I've observed in video retention charts, you're right in that there's a significant drop in views right at the 5-6 second mark. This happens practically 99% of the time with any videos I post above 15 seconds. But videos do give you a quick overview of a post if it's sized down to a bite-size clip and/or explains itself in the first few seconds after playback. I also don't think people's goals as they read are to efficient and quick. I don't think feeling slowed down is an issue for someone leisurely reading content. I work mainly with content meant for entertainment, but even for instructional or informative posts I think people want to enjoy the content they read, and doing so at a comfortable pace with visual content helps to make the learning process less boring and helps you retain more information.
Great post, chenry! What I find is most effective is the content itself and charts/videos to supplement it. This makes it bookmark-worthy. But it's also the listed points of take-aways that give me something I can use NOW, and make it stand out from all the other articles I bookmarked. It has a higher chance of me remembering it when I go back and use it to read it in depth and reference it for others.And sometimes a take-away can be quoted when you tweet the article! The take-away lures the reader in because they already found your post useful.
Absolutely superb data extraction. Thankyou
Looks like we'll have to brush up on video skills - and invest in more frequent haircuts as a result.
I agree, looks like video is the way forward. Just Hair cuts?! Why not wigs? help stimulate the economy a bit more?! :)
This is really remarkable work, and I wish we could say that internally we've done more of it (might get Sam on the case here soon). I'd be fascinated to see this kind of data analysis alongside traffic and with other vectors like published time/date/day-of-week.
Can't wait for part II :-)
Give me access to the MozVault and I'll get right on it. =) Part II will be done soon!
Big thumb up for you, the best post I've seen for this week.
I think it's definitely worth putting up in the main blog to show everyone else your fabulous work.
Nice post! As Rand always says, correlation doesn't prove causation .. but I accept your takeaways. Focusing on writing good long posts with visual content will help it become linkworthy. My nature is to write long and I've been trying to shorten because that's the advice I'd received, so this is helpful to me. Thanks!
I think you have to take into consideration whom your writting for when deciding whether to "write long" or "keep it short". For example at YouMoz a long detailed and graphically rich post seems to be the way. However, on another blog (which it seems we'll find out in part II) it may be the exact opposite.
So I guess what I'm saying is, do as the site/blog/forum dictates. If short is what everyone else is doing and the site tends to appreciate that type of post then go short. On YouMoz it seems long and detailed is the key.
I agree with you here. Know your audience when you are writing. Us SEO/SEM's tend to be long winded, write, and enjoy reading long post. Just remember this data is only based on SEOmoz and there will be more to come.
Thanks guys, good points. I need to stick with a long-winded crowd or change my ways :-) -- can you find me a market where people like long posts with no real point and no visuals? That's what comes most naturally. :-)
Aren't those called gossip columns!
Man, you did some hard work, especially with all the graphs! Big props to you! ;-)
Where did you originally read that blogs posts should be kept to max. 500 words? I think that's a nonsense rule, which you prove with the ILD stats to very large posts/pages - and therefore it's a good thing that you've solved a myth using quantitative data.
I've read that you should keep them around 500 words also. That was quite a while ago though. Interesting to see though.
Nice post. Re super headline from word cloud: How 'bout this beauty?
"New Matt Cutts advice for beginners: paid link SEO tactics & duplicate content practices are best value optimization strategies for Google search rankings!"
LOL
Amazing! That would generate links for sure!!!
great post, so this article im writing on google needs to be around 1800-3000 words and i can be famous...
I do agree more posts need to use visual content, and wow great images in this post, thank you for an easy post to read and understand.
Great post! Liked the insides. Thanks for sharing.
"Content is most important thing to a posts but posts with extra visual content attract extra links."
Content, as a category, covers a lot of ground and doesn't indicate level of information quality or value.
Any suggestions for defining content more specifically?
BTW, great post and thanks for all that work.
Damn Casey, way to take it to the next level. Makes my post look like dog crap! ;)
Seriously though, killer interesting data and great execution on the presentation.
You inspire greatest my man! Thumbs up to you.
Love that you normalized this. Thanks.
Looks like its time to write a long post with three types of media involved...could (and should be!) be interesting...
I read only the takeaways part of this post and I totally agree based on experience that posts accompanied by video and photos and other images receive not just more IN LINKS but also get longer sessions from viewers.
Nice post! Keep it up!
Interesting post! How about writing up a in-depth process using what' you've already done as an example, for people interested in running a similar test on their own blog?
Like some people have mentioned above, it varies from blog to blog and what kind of audience it has, so teaching people how to do this would make a great post.
Ouch - I hate making (or being in) videos. However we spend a fair amount of time cutting posts down to size (i.e down to 600-800 words).
My take away from this is that we can write more freely and get more links.
Hallelujah
No matter how much we wish and hope one thing remains a constant since inception: We Like to Watch TV
Very insightful post! I love. You must have spend a lot of time working on this post.
Wow great article, I like the take away points.
Great post. I had a question about the chart labeled: "Avg # of In Linking Domains vs Post Lenght". Is the data point labeled "3733 - 3035" in reverse order or is it a typo? Should it read "2733 - 3035"?
That would be a typo! Nice catch! It should read 2733-2035 like you said, I'll work on getting that changed!
Very common of me to have a typo in my comment about a typo. Lenght? :-) (Thanks, Jen, for pointing that out! haha!)
Yea that gave me a giggle. :) The image is updated btw!
Should the age (how old they are?) of posts within the categories have been analysed?
Older categories may have a majority of posts that are older compared to new categories with only newer posts.
These older posts might have more ILDs due to time live?
Sounding repitive...nice post and appreciate the work.
To often we get caught up with post length. It really depends on the audience and topic. An old english professor told me: "the length should be like a skirt...long enough to cover the important parts, yet short enough to keep it interesting."
Text vs. video, prefrence depends on topic, IMHO
Looking forward to part deux.
For a followup post you might also want to look at paragraph size. In an effort to convince the editorial staff to use shorter paragraphs I looked at average time spent on page vs average lines per paragraph.
It wasn't statistically sound, but basically we had many contributors that would write massive articles with massive paragraphs - I found that our visitors would actually spend less time on these than on far shorter articles with liberally spaced paragraphs.
I'd be willing to bet that would factor into the link aspect too.
Great information, thank you for sharing!
Beck @ https://profitseo.com
Looks like numbers sell, but they stay with you for the feeling. Great tips and guidelines you share here!
Something I've learned recently is that infographics are a class of images themselves, similar to GIFs. Infographics use a healthy mix of bite-sized text with an aesthetically pleasing display. Honestly, they may be the most optimal type of media for presenting a lot of different facts/data in a visually appealing way. I wonder what kind of data would appear when aggregating the stats from a content source in the posts with infographics vs posts without them.
Great informative post! I'm very curious about the bigger project you're talking about!
Congrats for making the main blog chenry. It's a great post and totally worthy.
"Posts with between 1800 and 3000 words will attract more than 15 times more ILDs than a post with less than 600 words."
Is this correct? The "Av # Linking Domains vs Post Length" chart indicates that posts with between 305 and 607 words attract 4 ILDs. Posts with 1822 to 2125 attract 19 ILDs. That's 5x more and not 15x more ILds. Am I misunderstanding something? It does indeed look as if there is a correlation between length of article and # ILDs. However, from an SEO perspective, is it better to get 6 ILDs for 600 words (to 3 different posts/pages on your site) or 18 ILDs to 1,800 words. I would have thought it better to have 6 links to each of 3 different posts.
Very interesting post. Love the recommendation for more visual stuff.
Great post. I have reviewed it several times. What tool(s) did you use to build the word cloud chart?
BTW, I think the 5 ILD note on the "Avg # In Linking Domains vs Media Type" chart should read 3 ILD.
Hey gbh, thanks for the feedback. I created a custom php script that generated an html cloud for me. I then took that and made the image with photoshop.
Posts like this makes it worth to be a pro.
Thanks
Very interesting post. I have / would be curious regarding length of time spent of page verses the number of links verses the number of media types per post. I know personally I would much rather read a lengthy post that contains graphics then one that doesn't and I would much rather link to a post that has graphic elements than one that doesn't most likely because I like graphics as they can help explain quickly what the post is about.
Fantastic, in-depth work you've done here, some really fascinating numbers are being displayed which will definitely be kept in mind henceforth. Thanks!
Good post, very link-worthy!
Thanks for the post, it was really inciteful! When I release my new website over the next few weeks, I'm going to be sorting out a half-decent camera so I can do a few videos online. This is where my posts are definitely lacking and I think it'll be good for me to use some video!!
i appreciate the work you've put into all the graphs and stats.
love the take-aways, short and to the point, definitely good guidelines.
looking forward for p2.
Fantastic post, but I have to agree with some of the comments above. It does entirely depend on the audience you are writing for. I prefer short articles that are to the point and include a graph/image or two to support the findings.
I don't always have time to read a 1800+ word article even if it is great content, so I might just flick over the highlights. I'm not likely to link to posts from my own site that are so long winded because I don't anticipate my readers will want to read such a long post.
Also I am not surprised around the video increasing the volume of ILD's linking in because the video allows for a user to watch, listen, and learn rather than having to read and scroll. We are lazy in nature after all! Not to mention you can say a lot more in speach and body language than you can in text. People often mis-understand what is said in a written format.
Rands WhiteBoard fridays are an excellent example of link bait.
I think the most important thing is the content as always. I also personally enjoy the charts, graphs and picture posts.
This is great and interesting that we also see less value in posts that get too long.
This is a really useful post! And it has colourful graphs to help summarise the data, which is always a plus point with me. I really agree with the multi-media approach, they always attract me more than simple text-only or video-only blog posts. I prefer text/image, with a supporting video. Time constraints mean that if the text and supporting images/graphs are good, then I'll bookmark a post and come back to it when there's time to sit and watch.
Thumbs up from me!
Excellent post, both in-depth and with actionable tips, thumbs up!
interesting stuff. i too am looking forward to part 2. i know for myself, i come to seomoz as a destination and know i will likely spend some time here. i often linger here and read all the comments whereas on other blogs i might just skim and leave. so i wonder if doing this analysis on other industry blogs will result in the shorter posts being more popular.
Very insightful as always. Do you mind if I ask what software you use to create those amazing charts?
The first two charts were built with a program call Fusion Charts and the other charts were created with Excel 2007.
Very well done. I will have to check out fusion charts.
My hypothesis would be that the media types graph is probably fairly universal for most blogs. I believe the length is highly dependant on the audience of the blog.
For example, a writers blog likely will attract people who like to read and thus longer posts may be good. (Do SEO's like to write?) On the other hand, a film makers blog might be better off with mostly pictures & videos and less words.
Hey, I love SEOmoz and I thought that these last days the articles were less interesting than they usually which I must say surprised me since it never happened before IMO.
I'm glad it changed with this post that is interesting, the follow up article you talk about at the end of this one seems very promising.
Grats and keep up the good work.
The data hurt my head this early in the morning! Love this post and the takeaways at the end are great because I didn't feel like looking at the data for too long
Great post. I like the way you tracked and normalized the data. All of your findings make sense too. I always felt that longer posts got more respect even if they weren't neccesserily as interesting or helpful.
awesome visual bar graphs! what program was used to make them?
Nice post chenry. I enjoyed reading the article, very helpful information and thankd for the chart, it gives me new insights, am looking forward for part 2.
your data seems like you spent some time on this thanks for the analysis i think its very useful and a great post