3 experiments on how to grow your blog!
This paper aims to outline the insights that were gained from three experiments conducted on the Tech2Classroom blog. The blog was initiated ten weeks ago with the mission to “become the go-to source for teachers and tech enthusiasts to stay posted on what’s hot in ed-tech”. Trough the blog we posted 18 articles on the intersection of education and technology, targeted at teachers, professors and technology enthusiasts. Having reached 1,740 viewers after eight weeks by promoting content through our personal social media channels, we decided to conduct three experiments. The data gathered through the experiments was captured by google analytics and the Blogger analytics tool. The insights gained from the experiments together with those results of other groups serve at a later stage to create a growth strategy for the blog.
Experiment 1 - Fake news
Our first experiment was to create a fake news article and spreading it through social media. Our working hypothesis was that spreading sensational fake news over the internet can attract a high number of clicks and lead to an outrage in the physical world. This experiment is inspired by the current role of fake news in the political context and its impact on public opinion. We started by creating a blog post on a topic that would be sensational and controversial, capturing it with a headline that a large group of people would click. As we were spreading the news in the IE environment, we decided to go for “Merger between IE University and ESADE Business School to be announced tomorrow”. The article was written in a way that people would think the news could be real, yet on the brink of being unrealistic to get the discussion started. Every group member shared the article on social media and in WhatsApp groups. As a result, 3,100 people clicked the original article in the first 6 hours before we changed the title to clearly indicate that it was part of a fake news experiment and deleted it from our social media channels. Even after the title was changed, another 1,200 people clicked the article. At this point, the group was contacted by 28 people, including the marketing department of IE asking for clarification. The article had been spread in Facebook groups and WhatsApp chats within the IE and ESADE communities. A group of bachelor students was furious, complaining that after the alleged merger they would have to move to Barcelona. The total amount of clicks generated was 4,400, making up more than half of the viewers of our blog. Of these users, 29% accessed the article from Madrid while 23% accessed from Barcelona. We are certain that this number would have further increased if we had not revealed the true nature of the article after 6 hours. At 81% however, the bounce rate of the article was ten percent higher compared to past articles. Additionally, the average sessions per user decreased from 1.65 to 1.27. We conclude that spreading fake news is a viable strategy for short-term traffic generation. However, as the interactions on the site were low, it is unlikely that a large number of readers would be transformed into recurring visitors. Therefore, spreading fake news is not useful as a long-term growth strategy.
Experiment 2 - Language adaptation for Taiwanese audience
Our hypothesis for the second experiment was that we would be able to attract even more Taiwanese readers if we published our articles in Chinese compared to publishing in English.
One of our group members is well connected in the Taiwanese social media space and shared our articles through her social media channels. As a consequence, 81% of our total readership before the start of the experiments was from Taiwan. For the experiment we published an article in English and Chinese. The European team members promoted the English article, while the Taiwanese group member shared the Chinese one through her Taiwanese channels. The English article was meant to a controlling measure. If the number of clicks of the European users had significantly varied from the portion of European readers on former articles, the topic of the article would likely have also distorted the number of viewer of the Chinese article. In that case, the findings of the experiment would not have been meaningful. After one week, the English article had reached 28 views, which was in line with the European share of viewers on former articles. In the same period, the Chinese article had reached 491 views. The range of Taiwanese viewers on past articles had been between 380 and 550. This shows that our current viewers from Taiwan do not have a language preference that would be meaningful enough to influence the decision on whether to click an article or not. We conclude that this is due to the fact that most Taiwanese viewers in our group member’s network have been educated in international schools and universities and therefore are very familiar with the English language. As a result of the experiment, we decided to keep on publishing articles in English language. To make sure that no external factors influenced the results, we need to repeat the experiment several times going forward.
Experiment 3 – Paid advertisement on LinkedIn
Through our third experiment we want to prove two hypotheses. Firstly, we wanted to show that paid advertising is a useful tool to attract users to visit the blog. Secondly, we wanted to prove that teachers and professors are more likely to click our articles compared to general technology enthusiasts, the two segments that we were currently targeting. The second component of the experiment is meant for us to get to know our target group better and to tailor our content accordingly. To conduct the experiment, we created a simple banner ad on LinkedIn, which is considered to be the cheapest form of advertising. The first campaign was targeted specifically to those users with a current occupation in teaching. The second campaign was targeted to technology enthusiasts. The two campaigns were supposed to take place on two consecutive days for 24 hours each. We started with the first campaign with a bid of €25. After 24 hours, the ad had been shown to 1,825 LinkedIn users (impressions). Of these users only three clicked the article, resulting in a cost per click (CPC) of €8.33. As blogger does not support the LinkedIn analytics plugin, we could not track the movements of those users on the blog. As the response to the first campaign was less than satisfying, we could clearly disprove our hypothesis that ads on LinkedIn are a helpful tool to attract readers to our blog. After the result of the first campaign, we did not start the second one, as we did not expect a better outcome.
The experiment was limited by the budget that we were willing to spend. To get more meaningful insights, we could have tried other ad formats and designs, increased the length of the experiment and could have bought ads from other social networks, e.g., Facebook, Instagram or Twitter. This would have likely led to a significant decrease in our CPC. Online research from other blogs suggests that a decrease of the CPC below €1.00 is achievable, making it more economical1. However, from the experiment we can clearly conclude that attracting followers through paid advertising is not economical in our context. Rather, we decided to keep on using our personal social media accounts to spread content.
From the experiments we learned that fake news are a great tool to drive short-term traffic to the blog, however it is not a tool for sustainable growth. From the language adaptation experiment we conclude that we will continue posting articles in English instead of translating our articles to Chinese. Finally, we learned that paid advertising is not an economically viable tool to grow our blog. Rather, we will continue promoting content through our personal social media channels. Apart from the experiments we learned not to violate copyrights as one of the group members was sued for €17,000 for copyright violations, charges that were later dropped. Furthermore, we experienced that people were more likely to click articles with catchy headlines and original content (e.g., opinion pieces). Last but not least, we learned that views are only one of many indicators to track. Others include interactions, comments, reposts, subscribers and session duration. We believe these findings are insightful and could potentially help other bloggers growing their blogs. Therefore, we decided to share the results of the experiment on our blog.
Experiment 1 - Fake news
Our first experiment was to create a fake news article and spreading it through social media. Our working hypothesis was that spreading sensational fake news over the internet can attract a high number of clicks and lead to an outrage in the physical world. This experiment is inspired by the current role of fake news in the political context and its impact on public opinion. We started by creating a blog post on a topic that would be sensational and controversial, capturing it with a headline that a large group of people would click. As we were spreading the news in the IE environment, we decided to go for “Merger between IE University and ESADE Business School to be announced tomorrow”. The article was written in a way that people would think the news could be real, yet on the brink of being unrealistic to get the discussion started. Every group member shared the article on social media and in WhatsApp groups. As a result, 3,100 people clicked the original article in the first 6 hours before we changed the title to clearly indicate that it was part of a fake news experiment and deleted it from our social media channels. Even after the title was changed, another 1,200 people clicked the article. At this point, the group was contacted by 28 people, including the marketing department of IE asking for clarification. The article had been spread in Facebook groups and WhatsApp chats within the IE and ESADE communities. A group of bachelor students was furious, complaining that after the alleged merger they would have to move to Barcelona. The total amount of clicks generated was 4,400, making up more than half of the viewers of our blog. Of these users, 29% accessed the article from Madrid while 23% accessed from Barcelona. We are certain that this number would have further increased if we had not revealed the true nature of the article after 6 hours. At 81% however, the bounce rate of the article was ten percent higher compared to past articles. Additionally, the average sessions per user decreased from 1.65 to 1.27. We conclude that spreading fake news is a viable strategy for short-term traffic generation. However, as the interactions on the site were low, it is unlikely that a large number of readers would be transformed into recurring visitors. Therefore, spreading fake news is not useful as a long-term growth strategy.
Experiment 2 - Language adaptation for Taiwanese audience
Our hypothesis for the second experiment was that we would be able to attract even more Taiwanese readers if we published our articles in Chinese compared to publishing in English.
One of our group members is well connected in the Taiwanese social media space and shared our articles through her social media channels. As a consequence, 81% of our total readership before the start of the experiments was from Taiwan. For the experiment we published an article in English and Chinese. The European team members promoted the English article, while the Taiwanese group member shared the Chinese one through her Taiwanese channels. The English article was meant to a controlling measure. If the number of clicks of the European users had significantly varied from the portion of European readers on former articles, the topic of the article would likely have also distorted the number of viewer of the Chinese article. In that case, the findings of the experiment would not have been meaningful. After one week, the English article had reached 28 views, which was in line with the European share of viewers on former articles. In the same period, the Chinese article had reached 491 views. The range of Taiwanese viewers on past articles had been between 380 and 550. This shows that our current viewers from Taiwan do not have a language preference that would be meaningful enough to influence the decision on whether to click an article or not. We conclude that this is due to the fact that most Taiwanese viewers in our group member’s network have been educated in international schools and universities and therefore are very familiar with the English language. As a result of the experiment, we decided to keep on publishing articles in English language. To make sure that no external factors influenced the results, we need to repeat the experiment several times going forward.
Experiment 3 – Paid advertisement on LinkedIn
Through our third experiment we want to prove two hypotheses. Firstly, we wanted to show that paid advertising is a useful tool to attract users to visit the blog. Secondly, we wanted to prove that teachers and professors are more likely to click our articles compared to general technology enthusiasts, the two segments that we were currently targeting. The second component of the experiment is meant for us to get to know our target group better and to tailor our content accordingly. To conduct the experiment, we created a simple banner ad on LinkedIn, which is considered to be the cheapest form of advertising. The first campaign was targeted specifically to those users with a current occupation in teaching. The second campaign was targeted to technology enthusiasts. The two campaigns were supposed to take place on two consecutive days for 24 hours each. We started with the first campaign with a bid of €25. After 24 hours, the ad had been shown to 1,825 LinkedIn users (impressions). Of these users only three clicked the article, resulting in a cost per click (CPC) of €8.33. As blogger does not support the LinkedIn analytics plugin, we could not track the movements of those users on the blog. As the response to the first campaign was less than satisfying, we could clearly disprove our hypothesis that ads on LinkedIn are a helpful tool to attract readers to our blog. After the result of the first campaign, we did not start the second one, as we did not expect a better outcome.
The experiment was limited by the budget that we were willing to spend. To get more meaningful insights, we could have tried other ad formats and designs, increased the length of the experiment and could have bought ads from other social networks, e.g., Facebook, Instagram or Twitter. This would have likely led to a significant decrease in our CPC. Online research from other blogs suggests that a decrease of the CPC below €1.00 is achievable, making it more economical1. However, from the experiment we can clearly conclude that attracting followers through paid advertising is not economical in our context. Rather, we decided to keep on using our personal social media accounts to spread content.
From the experiments we learned that fake news are a great tool to drive short-term traffic to the blog, however it is not a tool for sustainable growth. From the language adaptation experiment we conclude that we will continue posting articles in English instead of translating our articles to Chinese. Finally, we learned that paid advertising is not an economically viable tool to grow our blog. Rather, we will continue promoting content through our personal social media channels. Apart from the experiments we learned not to violate copyrights as one of the group members was sued for €17,000 for copyright violations, charges that were later dropped. Furthermore, we experienced that people were more likely to click articles with catchy headlines and original content (e.g., opinion pieces). Last but not least, we learned that views are only one of many indicators to track. Others include interactions, comments, reposts, subscribers and session duration. We believe these findings are insightful and could potentially help other bloggers growing their blogs. Therefore, we decided to share the results of the experiment on our blog.
Comments
Post a Comment