Using Zemanta to Find the Most Blogged-About Topics in Your Niche: A Case Study

Daniel Male is a search marketing manager for, an e-commerce company selling B2B products in the printing industry. Their main product lines are signs, tags, labels and mats, but they also offer a number of free services such as their bike registry and safety quizzes. Daniel uses Zemanta’s services to connect’s content with our network of bloggers and help with distribution and earning links from our bloggers.

I first heard about Zemanta from Ross Hudgens during his presentation at a Distilled conference in April. I learned many things there, but Zemanta was hands-down my largest takeaway. Distilled’s LinkLove conferences are supposed to inspire SEO ideas, which in my case, worked. Yet, besides just SEO, I walked away from the conference with something else entirely: a new method for content idea generation.

I work for SmartSign, an e-commerce company that sells a wide range of B2B products and spans across dozens of different industries, notably safety, traffic, construction, and transportation. Our products’ diversity creates opportunities for new content, but because our scope is so vast, it can be overwhelming to come up with ideas for fresh articles and blog posts that people will want to read (and ultimately link to). Our staff of writers uses a number of techniques to generate new content, but few initiatives have worked as well as looking through which Zemanta bloggers have linked to our articles.

Searching for Answers

I began by downloading my YTD recommendations report (June 7 – August 31st). What started out as general assessment of how my campaign was performing turned into an eye-opening realization that a lot more was going on. After poring over nearly three months of data – 30,607 recommendations on 233 different articles and blog posts – I began to notice some trends. A few of my articles had many times more recommendations than others. Other articles were acquiring lots of links, despite a scarcity of recommendations. Most striking though, is that many of the best-performing articles were all centered on similar topics. Since our content and research covers such a wide range of topics, I figured that the recommendations and generated links would be spread more or less evenly across our articles. I couldn’t have been more wrong. Our performance was extremely top-heavy.

I had hundreds of URLs to work with in my effort to find what our niche’s most popular topics were, but not much else. To make things simpler, I assigned a single keyword to each article, “replacing” the URL. In the example below, I assigned my article on bike signage a keyword, or topic, called “biking”.

Note: Many of the articles had multiple themes to them. This particular post is about signage, bikes, and law. However, this article wasn’t acquiring links because of its sign-related content. It was clear that my biking article was about bikes and was being recommended and linked to by bloggers writing about bikes. I used this same, albeit subjective, thought process to assign topics to all the URLs.

After assigning each of the 232 articles a keyword/topic, I was left with 67 unique ones, 45 of which had yielded links. I was then able to sort my data by the number of links acquired, the number of recommendations, and the link conversion rate (“LCR”). Here’s what I found.


Caveat: All of the information in this post including the examples is true, but for the sake of my company I am intentionally excluding much of our data. I don’t want to give away all of our secrets!

SmartSign’s Top 10 most RECOMMENDED topics:

Takeaway: These topics seem to be the most blogged about topics on the blogosphere that relate to SmartSign’s niches. Who knew that out of all the topics that pertain to our company, biking, signs, and OSHA were the most popular? Only by consolidating our content into broader topics could I have learned this.

In fairness, Zemanta just makes recommendations to its users. The case could be made that there are many bloggers in relating niches that don’t use it and would therefore not have the opportunity to link to our articles. Yet, we must remember that there are over 140,000 Zemanta users and they probably cover most topics out there. Also, bear in mind that these top 10 recommended topics are not necessarily the ones that have generated the most links. If I wanted to simply increase my recommendations, without also focusing on linkability, I could create content centered on these topics. However, as SEOs know (and are often shy to admit), impressions and time-on-site is good, but it’s really the links that matter.

SmartSign’s Top 10 most LINKED-TO topics are:

Takeaway: This is, of course, the metric that I really care about. These topics (mostly) represent the best combination of most blogged-about topics, accurate recommendations, and high-quality content. Like the previous list, this one helps me figure out which topics are most popular, but when comparing it to the other two charts, I can determine which of our articles are most linkable.

An interesting discovery is that “drowning”, or swimming pool safety, has unexpectedly had thousands of recommendations, yet has only yielded 4 links. This leads me to believe that although there are a lot of bloggers out there who talk about swimming and water safety, our related content that covers the topic of drowning is probably low quality or irrelevant to what other bloggers care about. If I know this topic will get a lot of recommendations, I can now try to tap into the swimming pool safety “blogging market” with more appropriately targeted content. The recommendations are there; right now I just need to get my links up.

SmartSign’s Top 10 highest LINK CONVERSION RATES are:

Takeaway: This chart shows topics that have yielded the most links with the fewest recommendations. This is a reflection of accurate recommendations and, mainly, high quality and original content. You will note that all of these have few recommendations, yet have still been acquiring links. The posts that fall under these topics are worth emulating in tone, writing style, photos, and keyword placement.

Note: I have only included topics with multiple links to factor for “flukes.” There are a number of very specific articles that have had higher LCRs than many of the ones above, but they have only yielded one link. An article we produced on asbestos falls into this category. It is high-quality, but it’s too specific of a topic, with not enough data (one link on 20 recommendations) to be significant for this study.

Here are some other interesting gems I found from my data set:

  • Out of 45 total link sources, (a/k/a topics,) just three of them, or 6.6% (“osha”, “biking”, and “signs”) were responsible for 34.4% of the links.
  • Our total campaign Link Conversation Rate is .4% (125/30,607), which seems, at a glance, to be poor – but in actuality is quite good.
  • People really like talking about dog poop (899 recommendations).

Applying the Changes After learning that there are hundreds (if not thousands) of bloggers out there that write about dog poop, OSHA, biking, traffic, distracted driving, and water safety, our team began focusing more content around these topics. It forced us to think of new content ideas that biking blogs would be interested in. We now track OSHA violations and the latest legal interpretations. We include pop-culture themes and interesting facts in our articles about signage and regulations.

While writing, we include all the popular keywords under a given topic for each post. So, if I decide to write about bike lanes in New York City, with the umbrella topic of biking, I try to include as many of the bike related keywords as I can in the article (as well as in the URL and meta tags) – “bikes”, “bicycles”, “cycling” etc. – without keyword stuffing and sacrificing quality. That way, when a Zemanta blogger highlights one of the keywords in his post, our article has a greater likelihood of being recommended.

(Zemanta Note: our recommendations are based on more than keyword matching. Zemanta can also judge relevancy of different topics mentioned inside the same text and put emphasis on the one deemed most important when recommending content. Our engine understands when somebody is speaking about London in context of Ontario, Canada or England and recommends appropriately. Recommendations are therefore more precise. The background knowledge comes from big databases and Zemanta’s consistent learning.)

From glancing at all of my data, I also learned that bloggers really like current events and newsy items. Most of our linked-to content has been from SmartSign blog posts about things like mining accidents, a new bike registry, or a human interest story.

Here’s a good example of how we unknowingly (before this case study) used this to our advantage. Aside from being potential fans, our main product lines have little to do with the Brooklyn Nets. But we were able to tie them into a post about how their logo was inspired by old subway signage, a highly relatable topic. Our content was now being recommended to Brooklyn and Brooklyn Nets blogs which ordinarily would have no reason to link to us. Focusing on a popular topic such as the Nets generated three quality links to our blog. It’s worth it to make a point of merging something from our industry with a mainstream, popular topic for future pieces of content.

How YOU Can Get the Most Out of Zemanta

The changes SmartSign has implemented in its blogging strategy have led to a lower cost per earned link from a variety of authoritative bloggers. In case you would like to create a similar study of your articles campaign, here are the steps to do so.

  1. Download your recommendations report, and open it up as an .xls file.
  1. Go through all of your URLs and assign each of them a single keyword or topic. This may take a while depending on how many articles you have indexed. You can do a shortened version of this by discounting any articles that have no significant link or recommendation figures. However, if you go this route, your data will be less accurate because you aren’t using everything you’ve got.
  1. Once you’ve assigned each article a topic, merge the data (recommendations, links, and LCRs) for all of the duplicate topics, so that each topic has only one line.
  1. Then, make 3 separate charts and sort them from highest to lowest for recommendations, links, and LCRs.
  1. Begin searching for trends and outliers. Once you have your data set up correctly, the rest comes easy. You will find things that you weren’t even looking for and discover answers to questions that you never asked.

I’m sure that other users have found their own creative ways to use Zemanta. If you have, or have any questions about my processes, I’d love to hear from you. Leave a comment, or drop me a note by way of twitter or e-mail. And now, a cheesy sign cartoon:

If you are interested in using Zemanta to promote your content using our network of 140,000+ bloggers, please get in touch with us.

Enhanced by Zemanta

Subscribe to newsletter