Lehigh University: Increasing search referrals through keyword placement in article metadata
The Brown and White is Lehigh University’s student newspaper. We publish twice a week in print and constantly on our website at TheBrownAndWhite.com. For the past two years, The Brown and White has used Parse.ly in order to track the performance of our stories.
We have been able to use Parse.ly data to help us find a keyword that boosts visibility in search engines.
As we have been paying more attention to our analytics and tailoring our approach based on these metrics, our pageviews and visitors have steadily increased in numbers. For the past few semesters, we have been using Parse.ly to keep an eye on what can help our content reach more readers.
This fall semester, we decided to take a look at using the word “Lehigh” in various places within our article metadata to improve searchability. In the past, our mindset had focused primarily on pushing our content through social channels. But this semester we took some time to consider the audience we may be missing who might be looking for information on our school through search engines.
The research question we investigated this semester was: Are we getting the most out of our search referrals?
Our initial answer to this question was, “No.” The Brown and White had not used any deliberate strategy to increase search referrals, so it follows that more can be done to improve this metric.
So, we decided to try to find ways to tap into this potential audience to see what would work, and to identify what people might be searching for that would bring them into contact with our articles.
One idea discussed during the fall 2016 semester was to include the word “Lehigh” in article metadata to investigate where its placement could best increase the number of search referrals.
This was tested in two ways: comparing articles that contained “Lehigh” in their headline to those that did not, and comparing articles that contained “Lehigh” in their URL with those that did not.
For both analyses, the first step was to generate the data. A report was created on Parse.ly consisting of all posts from this semester, which began Aug. 29. Because some of the posts in the report had been written before this starting date, the report was adjusted to include only posts that had had been written this semester.
The data was then cleaned, removing outliers that might influence the results. All of the posts above the 95th percentile in search referrals were removed to avoid skewing results. With this cleaning, the top post left in the dataset had 137 search referrals, and 430 of the original 453 posts remained.
Additional variables were then added to the dataset, ones that indicated whether or not the word “Lehigh” was present in the headline of the article and in its URL. Only 196 of the 430 articles had “Lehigh” in their URL. Even fewer had “Lehigh” in their headline, with only 163 containing the word.
Results and discussion¶
The first comparison investigated whether putting “Lehigh” in the URL would help to increase the number of search referrals. The results, broken down by section below, seem support the initial research idea.
Three of the five sections shown indicate an increase in search referrals when “Lehigh” is included in the URL of the article. Lifestyle, multimedia and sports posts all show increases of at least 15 percent and as much as 40 percent in terms of search referrals when “Lehigh” is included over when it isn’t. This is an easy change to make for most articles — “Lehigh” can be easily added to an article’s URL without much effort. This little change could provide enormous benefits for a post’s search referrals.
Meanwhile, the fact that news articles without “Lehigh” in their URL received more search referrals than those with it is not as concerning as it initially seems. A two-sample t-test reveals that the difference between the two amounts of search referrals is actually statistically insignificant at the 5 percent level; that is, there is greater than a 5 percent chance that the difference in the amounts of search referrals is due to chance.
Opinion posts, too, present less of a concern than they seem. Many opinion articles don’t pertain directly to Lehigh’s affairs, and it would make little sense to include “Lehigh” in the URL of these posts in an attempt to increase their search referrals. Since the emphasis of this report is to maximize search referrals, including “Lehigh” in posts that don’t concern the school would be a pointless exercise and therefore the opinion section can be largely ignored for the purpose of this analysis.
If including “Lehigh” in the URL of a post suggests an increase in search referrals, including it in the title offers more conclusive evidence:
It has been shown that including words, specifically “Lehigh,” in a post’s metadata can help increase search referrals.
Here, four of the five sections show an increase in the number of search referrals when including “Lehigh” in the title. The sports section shows nearly twice as many in search referrals with the inclusion, while lifestyle and multimedia again show significant increases, as well. News, while now showing limited benefit, still remains inconclusive on whether or not including “Lehigh” helps, but it now sits in the positive category rather than the negative. Opinion continues to show a decrease in the number of search referrals, but as noted, the section can be largely ignored because the URLs and titles for its articles are already appropriate.
Regardless of whether or not Lehigh is in the URL, it appears that on average, articles with Lehigh in their title receive more search referrals than do articles without Lehigh in their title. “Lehigh” in the URL seems to have less of an impact on the difference in search referrals, showing a positive difference when “Lehigh” is in the title and a decrease when it is not. However, it’s easy to see that the most search referrals occur when “Lehigh” is in both the title and the URL.
It has been shown that including words, specifically “Lehigh,” in a post’s metadata can help increase search referrals. This is a simple approach that can lead to as much as a 100 percent increase in search referrals, as seen above. However, this is not the only approach that The Brown and White can take to increase its readership.
One other such way is to change the time that an article is published. Publishing articles at different times can influence site traffic and the number of views an article receives. Here, the hour in which the article was published was plotted against the average number of search referrals for posts published during that hour. Additionally, the total number of posts from that hour is indicated by the size of the point, showing which hours feature the most published posts.
This graph helps to show the pattern of search referrals that The Brown and White is receiving. The first obvious conclusion from this graph is that no posts are being published between 3 a.m. and 8:59 a.m. This seems intuitive, as it’s when the majority of the intended audience is sleeping. However, the late morning hours are when a lot of students wake up for class. Having no posts before 9 a.m. and only 2 before 10 a.m. means that a potentially good time for search referrals is being missed. Without more data, though, it’s difficult to determine how successful this period would be.
Through considering search referrals, we have been able to use Parse.ly data to help us find a keyword that boosts visibility in search engines.
Most of the search referrals, oddly enough, come during the late night and early morning. The highest mark actually comes during the 2 a.m. hour, although this may not be statistically significant because of the small number of articles published then. Midnight also seems to be a successful time for posts getting search referrals with a combination of a large number of posts as well as many search referrals per post.
The other best time, which The Brown and White is already taking advantage of, is the evening and early night hours. Two of the three highest marks for search referrals during the 5 p.m. and 7 p.m. hours, which also happen to be times when many articles are published.
In general, it seems like the best times for search referrals are already being utilized, and apart from the late morning hours, other times have already been tested and proven to be less beneficial.
Conclusion and takeaways¶
Overall, adjusting the way we were looking at our metadata and SEO helped garner more pageviews through search referrals. It is an effective method of increasing readership through channels we had not previously targeted.
This is a step we have added to our SEO process to try to include “Lehigh” as much as possible, when it is relevant to do so. We have seen it impact our views and help tap into a new group of readers. Previously, we had started to plateau in terms of getting the same visitors to come back through social channels.
But now through considering search referrals, we have been able to use Parse.ly data to help us find a keyword that boosts visibility in search engines. It’s fairly simple, too. We are the newspaper of Lehigh University, so our content is likely going to be relevant to anyone who is searching our school.
This discovery particularly emerged from our sports section where we realized that during a particularly good season, people would be searching for “Lehigh football.” Trying out this method gave our recaps and game previews many more views than they used to get.
This is something we can expand upon in following semesters, using the same method to try out other keywords in search referrals.
This, in turn, made us more mindful of how we structure search engine optimization strategies in posts — especially making sure to differentiate web headlines from what we usually write for print. This study helped us to increase our viewership and consider a new way to garner views.
Audience development guide
We gathered advice from audience development professionals and found three areas that everyone agreed you have to tackle to successfully grow and engage audiences.
The beginner's guide to audience analytics
A five lesson course that guides you through the five facets of implementing data into your workflow so you and your team can make better editorial decisions.