What really makes a blog post worth linking to? In my last post, What Makes a Link Worthy Post - Part 1, I took a look at the 3,800 blog posts on SEOmoz and did some analysis on a few different aspects of the posts and their affect on the number of in linking domains (ILDs). Some of the results were very interesting to me and it made me want to push it further.
I created a list of 40 SEO/SEM blogs that I read and feel are important to people in the industry and set those as my sample population. I first crawled each website and collected a list of over 72,330 different blog posts from the 40 different websites. Then over the course of the next few days, I crawled each post and collected the following information in my database:
- Blog Post Title
- Original URL
- # of Links from Root Domains (Via Linkscape API)
- # of ILDs (Via Linkscape API)
- If The Post Had Images, Lists, Or Videos
- Content of Post (No Comments or Other Text on Site)
- # of Words in Post
POSTS TITLE EFFECT ON ILDs
Does the length of the post’s title affect how many domains will link to it? The data suggests that posts with a title length between 10 and 18 words are on average more linked to than those with less or more. The data also suggests there may be a “sweet” spot around 14 to 16 words in length. The chart below was created without removing stop words.
This data proves to me that a descriptive title is what the linkerati is looking for. Going overboard on the length of the title can prove to be a bad move also.
EXAMPLES OF HIGHLY LINKED TO POSTS WITH TITLE LENGTH IN THE “SWEET SPOT”
- Google Says: Yes, You Can Still Sculpt PageRank. No You Can't Do It With Nofollow
- Big Brands? Google Brand Promotion: New Search Engine Rankings Place Heavy Emphasis on Branding
- What Makes a Good Web Directory, and Why Google Penalized Dozens of Bad Ones
POSTS LENGTH EFFECT ON ILDs
Post length is a long debated thing out there in the blogosphere. Most bloggers will tell you that you should keep your posts around 500 to 900 words, and that might be stretching it. When it comes to SEO/SEM blogs, longer more content filled posts are more linked to than those with limited amount of content.
From the chart below you can see there is a word range that seems to collect more ILDs than other word ranges. Based on the data, the ideal length of your posts should be around 2328 to 2618 words. In my previous post, the ideal length for only SEOmoz’s post was between 1800 and 3000 words.
The chart above shows posts only up to 2812 words, but accounts for over 99.55% of all the posts. Posts that were greater than 2812 words really had a low number of ILDs. For this reason and for the display of the chart, they were removed.
EXAMPLES OF HIGHLY LINKED TO POSTS (BETWEEN 2328 AND 2618 WORDS)
- 101 Ways to Build Link Popularity
- SMX East: Give It Up
- Five Link Development Experts: A Group Interview
DEPTH OF POSTS EFFECT ON ILDs
Seos know that you want to keep your key content in as few subfolders as possible but does this affect the number of ILDs you receive? The data suggests that the depth of your post doesn't affect the number ILDs. The graph below shows that just about half of the blogs out there place their content two subfolders deep, such as seomoz.org/blog/POST-TITLE.
MEDIA’S EFFECT ON ILDs
What role does placing list, images, and/or videos in a post play on the number of ILDs? The data shows that putting any one of the media’s in your post will increase the number of ILDs you receive. Putting a list on your plain text post could double the number of ILDs you receive. The results are even more outstanding when all three types of media are used.
Do I really believe that you can take any post, slap a picture in it and you will automatically receive more links? No, but if you have decent content and media to support your post, it will appeal to more users and in turn increase the number of potential links. I find it amazing that just by adding images and lists to your post could increase the number of ILDs by a large percent. Images and lists are one of the easiest things to create and anyone can do it, so why aren’t they? See the chart below for the full specs on adding media to your post.
TOP MEDIA POST EXAMPLES
So I’m sure you are all wondering what some good examples are of the different type of post along with the media. Below are some links to some great posts that contain different types of media and have been successful. Some of these posts should be your guide when creating new content for your site.
ALL 3 MEDIA TYPES
ONLY LISTS & VIDEOS
ONLY LISTS & IMAGES
- Big Brands? Google Brand Promotion: New Search Engine Rankings Place Heavy Emphasis on Branding
- Canonical URL Tag - The Most Important Advancement in SEO Practices Since Sitemaps
ONLY IMAGES & VIDEOS
ONLY LISTS
ONLY VIDEOS
ONLY IMAGES
NONE
TOP DOMAINS FOR MEDIA TYPE
The data shows that there were certain domains that tended to use certain types of media in their posts. Below I’ve put together two sites for each category so if you enjoy posts of a certain type you can visit their blog.
ALL 3 MEDIA TYPES
ONLY LISTS & VIDEOS
ONLY LISTS & IMAGES
ONLY IMAGES & VIDEOS
ONLY LISTS
ONLY VIDEOS
ONLY IMAGES
NONE
AUTHORITIES EFFECT ON ILDs
What role does a blog authority play in the number of ILDs? Seems like a simple question and the data seems to show that if your an authority in your niche, you will generate many more ILDs than someone who is not. Look at the chart below and you can see that Matt Cutt’s blog generates almost twice as many as its closest competitor, sugarrae.com!
TOP TOPIC THAT ATTRACT LINKS
Unlike SEOmoz not every blog places their post into nice categories and if they do, those categories will not match across all the sites. So how do we determine what topics are attracting the most links and are good topics to create posts about? We crawl 72,300 posts, determine the ILDs, and then extract the most used words from those posts to create a “super” group of keywords that result in link worthy blog posts.
The first thing I wanted to do was extract all the text and find the words that are most used in all blog posts, I was curious, aren’t you? After pulling out 27,658,728 million words and sorting them, five words came out on top: Search, Google, Yahoo, Site, and SEO. Was I surprised, no but it’s interesting to know and a good starting point.
Taking a look at the top 1% of all 72,330 posts, it was found that the words did change a little bit. Some of the top words used were: Search, Google, Site, Links, SEO, Content, People, and Social. This data seems very similar to what was found in part one of this study, with the SEOmoz data. Posts that are about link building are very popular but now we can conclude that they are attracting links. When we look at a much smaller percentage say only the top 50 posts, you find that you are getting very similar words such as: Google, Search, Blog, Link, Pagerank, and Site.
So what can you really take away from the content of the top 50 blog posts? Stick with the major engines: Google, Yahoo, and maybe even Bing, on a good day. The linkerati likes topics including Link Building, Pagerank, and Social Media. As my disclaimer stated above, these are not the rules but just observations from a small sampling of the blogosphere. If I knew the exact topic that the linkerati loves, I wouldn’t be writing here, I would be out making millions writing all day.
BIGGEST TAKEAWAYS
- The data suggests that posts with title between 14 and 16 words attract more ILDs than those with longer or shorter titles.
- Contrary to belief, the data suggests that posts with more than 900 words are attracting more links than those with 900 words. Shoot for post between 2328 and 2618 words.
- The data suggest the location/depth of your blog post doesn’t seem to have an effect on the number of ILDs you will receive but may affect your SEO work, so be cautious.
- If you’re interested in the top post with a certain type of media, check above. Also if you’re interested in the blogs that tailor to a certain type of media, check above.
- Authority plays a major role in the number of ILDs that you will receive on your post. Matt Cutt’s blog receives twice as many ILDs as the next closest blog.
- Hot topics that attract links include: Google, Search, Blogs, Link Building, Pagerank, SEO, and Social Media.
SUMMARY
In summary, the takeaways above are generalization about a small group of post from the blogosphere and should not be taken as rules but merely as a guide to help you create content that will have the possibility to generate links. Work on the authority in your niche and become that place people come to receive great advise. While you’re waiting for authority to grow, make sure that your posts included visual aids to help readers get the takeaways quickly.
SPECIAL THANKS
Special thanks to the SEOmoz team for the access to the Linkscape API. Without the use of the API this post would have never been possible.
a great followup post, and a thumbs up from me - but a lot of what your mention I fear could be put in rands favourite section of causation vs correlation...
for example: the number of words in a title, I would be worried about optimizing based on that - and likewise the number of post comments per blog overall, its not something that can be optimized towards really - but more of an indicator as to the strength of the original author.
MOGmartin
Well there is a huge difference between Rand's post and mine. Rand's had to do with your position in the SERPs where mine is just about the collection of links. If you use the takeaways from my post it doesn't mean you will have high rankings just that you have the potential for getting a high number of links.
Chenry, you've written a great follow-up post IMO. Big props to you!
I agree with you that the number of words in the body part of a post is a very good indicator of the popularity chances (number of ILDs) of that post. To me that's quite logical: the more words in one post, the more in-depth that post will probably be regarding the specific topic(s) it describes, and thus the more valuable it will be towards the audience.
Regarding the title length measured in words, I guess the 'sweet spot' (I would use 10 - 20 words as a guideline) is just a logical thing to do.Less than 4 words doesn't describe the page contents well, more than 25 words seems a bit spammy.
For future research, how about doing the same study on non-blogging sites about varying topics? I'm curious to see if the same outcomes apply.
I'm looking into doing aother post with different types of site, though I haven't started it yet. Any ideas would be helpful!
Looking forward to that specific follow-up already! ;-)
Fantastic post, but like MOGmartin mentiones, I would be careful about optimizing according to these studies.
The stuff about media effects however is really valuable to me. I run a photography tutorial website and will make more of an effort to include everything in our tutorials from now on!
Thanks for taking the time to put this stuff together!
Kenneth, great points and I agree with you. Like I say in the last paragraph, these are only observation and not rules, so use them wisely.
I agree with Kenneth also , Information must be taken into consideration but always should be passed through mind filter
I read through this with the words Causality Vs. Correlation in my mind.
With the exception of Authority and media I would discount most of these findings.
Fantastic post with some really good information that we can take away and play around with.
It would be very interesting to analyse aspects of site design in relation to the points you made to see if we can spot any correlations.
I also see you only had 1792 words on your post :P
Like I said, not rules to follow but just suggestions. =)
SOme pretty interesting stuff in here; I wasn't too surprised about more people linking to media rich posts but I was surprised about the length - it seems I have underestimated the attention span of my fellow man (and woman!).
I felt for a long time that in general, we as SEO's/SEM's tend to underestimate the attention span of our viewing public. I think that people heard or saw the stat that "everyone wants things quicker then quick on the internet" and thus it became dumbing things down.
If someone doesn't want to read your stuff they won't read it if it's 1 paragraph or 1 page. Those who are interested will read, as long as the information is presented in a clear, enjoyable fashion. (IMO).
More likely you've just underestimated the attention span of your fellow linker.
I am full of shame...
I'm off to write a massive blog post on this right now.
Somewhere around 2500 words?
Great research, and definitely some useful takeaways here. I think what it boils down to is that interesting and well formed posts that engage the user both visually and technically (lack of a better term) will tend to get more links/popularity.
I think you just need to write about what you know, what you are interested in and do that in such a way that is entertaining/interesting for the reader as well.
"The data suggests that posts with a title length between 10 and 18 words are on average more linked to than those with less or more. The data also suggests there may be a “sweet” spot around 14 to 16 words in length."
Wait, what about correlating this with the actual distribution of title word length? If there are relatively more titles having 10-18 words then it's not surprising that the sheer number of links to those posts is higher.
Yeah, I am wondering if this was normalized data or not...
Chenry,
Great post as is usually the case. I am curious as to the time frame that your through research covers... Months, Years or is not a factor. In todays Status Update, What I'm Doing and Tweet induced web, links are given with less frequency than as little as a year ago and I am more than a little curious as to how this will shift the web "linkscape". Maybe not a factor now, but if you do a follow-up in a year, there might be some identifable shifts???
Cheers!
James,
The research covers all posts from these 40 blogs. Much of them go back years, though I will admit I never looked into how time as an effect on some of these factors. Obviously the older sites have much more time to collect links but with the large amount of posts in the data set, I think they all work themselves out. I will look into it though.
Very nice, the title length details here is the most interesting thing for me. There is a definite balance between posts that aquire links and posts that within their own content are totally optimized in an "on site" perspective
An excellent follow up post, congratulations. I love the graphs by the way, they make me more likely to link ;-)
I do agree with the other comments however, that optimising on these areas doesn't guarantee improved links (though I realise you are more than aware of this yourself).
Ultimately if Matt Cutts wrote articles with 1 word titles, small posts of less than 200 words people would still follow him and link to his articles.
Its very subjective, but your analysis does provide some good high level pointers.
Thanks again.
that's is truly amazing work. How long did it take to compile.I'll bookmark, get a cup of coffee and read it again. Thanks for sharing this knowledge.
What tool did you use to create https://www.flickr.com/photos/21865686@N05/4051008829/ ? The best pie chart ever. Forever :)
chenry, your addiction to moz-points blows my mind. I fully encourage your addictive habits. This is awesome!
LOL. He is the penultimate mozpoints addict. Ever since Rebecca's moz points were revealed, it became an obsession with him.
Luckily, we're the beneficiaries of his addiction.
What a great article. Very informative, and I want to hand you over a huge thanks for publicshing all of this work. Very generous of you. Thanks again.
Great post, I especialy liked that you showed the correlation between your findings in part 1, and part 2. I think its interesting that they both line up pretty well.
For part 3 your could do all non seo blogs. I would be interested to know if the 900 word count actually holds any water.
Great take aways, thanks for all the effort.
Great data collection post, chenry, thanks a lot.
Hat off for your effort, last time was data from seomoz, this time is expanded to 40 blogs, WOW!
It was no surprise for me to see mattcutts is always the boss, but sugarrae, I will definitely pay more attention to her blog in the future.
Great takeaways, thanks.
Hi,
Wow ! Nice Article.
I'm shocked at the 14-16 word title findings. The first impression I had was "WTF? 14 words in a title?" I personally think that is WAY too wordy, and I have never really gone beyond 9-10 maximum. But, if it works, it works I suppose. Great article.
Pixelrage,
I was a little shocked at first too. This 14-16 word range includes those stop words which normally don't get placed in things such as title tags. I think the main take away should be to have good descriptive title.
Bummed Search Engine Land wasn't included in the list of blogs you did the survey against. I can't say it would have skewed the results much. We get lots of links to our content, and what you've summarized matches what my gut shows -- descriptive titles for substantial articles help.
What may skew your results is using only SEM blogs. What's linkworthy in general and what's linkworthy for SEM blogs may be two different things.
Danny great point there, each blog type is going to be different on what is linkworthy. I would of loved to have Search Engine Land in my data but I didn't have unrestricted access to all archives. Maybe in my next study I'll try to get access.
Very useful for expecially business and ecommerce sites for optimizing and getting goog position in the SERP major search engines.
you are giving new clue for getting more and more response by controlling the lenth of title of post.
mot effective title words lenth is 14 to 16 word.
Hi Casey,
Great post and great work. Question about number of words in title, for clarification:
This shows a correlation between X, the average number of ILDs, and Y, the number of words in the title. Was this a weighted average? Do we have a very large population of 14-16 word titles, or a very small one, conferring an anomalous mean if one or two of the posts in that specific population were major link-bait? (Also, in the over-25 word-title categories, I wonder if there are even more than a handful of posts with titles that long?) I'd love to see the distribution of number of posts X, with number of words in title Y.
The rough average of the number of links obtained per title-length doesn't tell us enough to know if there really is a sweet spot, because we aren't given enough insight on the size of the samples/populations.
Out of the 72K blog posts, I'd venture that the vast majority of posts had titles under 11 words in length. Some of those posts will get many links. Most will get few. If we then look at the smaller sample of 14-16 word titles, a few heavy-hitter outliers will skew the entire average.
Thanks for your help!
Chenry,
Thanks for all your posts. This one is especially helpful. You have really help me out with my site as well https://www.apartmentninjas.com.
Way to go chenry. I can now start measuring the time it takes for your YOUmoz posts to go mainblog in hours instead of days.
Congrats on another winner
Thanks Goodnewscowboy! I'm still waiting for your first YOUmoz post.
Yeah but you've so ruined my chances to ever meaningfully contribute with this latest addition to your already stellar postings.
You've raised the bar so high, I'll need to hire a chimney sweep to throw me a rope just to be able to reach it.
This research really took a lot of time to complete. This one is really useful and interesting. Great job and hope I can use this data in the near future. Thanks a lot.
Nice post and glad I made time to read. Takeaways are very helpful and the examples are money. As I am a novice, here are a few questions/comments.
-Won't a combination of the audience & topic really determine the ideal post length for obtaining ILD? You looked at 40 SEO/SEM blogs. Would the length sweet spot be the same for posts on the Detroit Lions or sports?
-You said: The graph below shows that just about half of the blogs out there place their content one subfolder deep. According to your graph it is 2 subfolders. Believe it is one. Looking for clarity.
Harlan,
Like myself and the others have stated. These observations are just for the 40 SEO/SEM blogs and could be different for other blog types. I would think that a sports blog might be a little on the shorter side depending on what the topic was.
I would take a look at other blog in your niche and use the SEOmoz toolbar to see what post are getting a high number of ILDs. Then take a look at the length of the post, media, and title length.
As for the subfolder depth, you were correct I did have a typo. Most of the blogs where in the 2nd level of subfolders.
One of the biggest concepts I've taken from leadership training is that our ability to influence others is rooted in emotion. Influencing someone to link out to you requires reaching them on a primal level. I believe that each of your biggest takeaways support this idea. Taking a psychological approach to link-building allows us to see how our feelings guide our behavior.
https://www.seomoz.org/blog/the-emotions-that-make-us-link
This is another home run, go to article in the SEOmoz library.
Thanks.
Thanks for this data. Incredibly useful as I'm still learning all this stuff. Your numbers don't lie, so I'll be much more carefule with my post titles in the future. I've been guilty of short titles in the past, but it kind of makes sense that a longer, more descriptive title is the way to go.
More likely you've just underestimated the attention span of your fellow linker.
Awesome post. I found the part about the 'title length sweet spot' and blog depth intresting. The graphs were really colorful and great illustrations of the stats you found. Nicely done.
I love that you are applying data to blog posting. So often advice on blog posts are more opinion-based. This really speaks to the data-driven SEO marketer : )
Amazing follow-up post. It's funny, this is basically the 'perfect post' according to your criteria.
Thanks for the pointer to sugarae. And it is not really surprising that Matt gets the links given who he is and what he does. Which makes me go with the cause and effect crowd for a lot of this - Cutts would still get the most links if his posts had one word titles, no graphics, video or pics and if they were only 100 words long.
But the general point is really well made and we will be making sure that we have a greater variety of media in our blogs going forward.
Thanks
Nice work here and very interesting data crunching indeed!
It'll be very nice if you can also include the number of pageviews for each articles.With the pageviews data, we can calculate number of pageviews per link generated.
I think this will be very a very good complement to your analysis to find out which factors can help to improve the % of conversion per pageview to the article.
Hey, what program you used for making the pictures on this post?
I used Excel 2007 for the graphs themselfs then added a little extra touching up in Photoshop.
love the post i find the media graph very interesting its so cool when you can see such apparent patterns
i totally knew this was going to be promoted
great work man cant wait to see some more posts by you