I’ve spent the past month analyzing an Urchin Report on the usage of Medart. Google Urchin was Google Analytics predecessor. I was using Urchin 7 which is the newest version, but it was released back in 2010, so there haven’t been any recent updates to it. I was warned that not every aspect of the software would work, and I found that to be true. There were no visitor data, but there were data on how many visits and hits there were to the site. Data on visit length did not appear consistently and neither did total bytes used. I was excited to find that Urchin had recorded bot activity, which gave me a more accurate depiction of site usage. We are more interested in use by real people, so I extracted bot hits from total hits to see how many real people used the site. However, Urchin states that it identifies bots by looking for “bot-like” activity which means there could be some inaccuracies. Therefore, while this data is helpful and shows some trends, it is possible that the information isn’t exact.
I only had access to usage data from the past 18 months. 18 months initially consisted of Sept. 2015 to Feb. 2017, but since it is now March I have lost access to the Sept. 2015 data. Luckily, I had already recorded everything I needed from Sept. 2015 before the data disappeared. It appears that Urchin can only store 18 months of data at a time.
Since I worked on this mostly in February, I’m going to focus this post on data from Sept. 2015 to Feb. 2017. When looking at total and average visits to medart, it looks like there is a slight low in the summer months and a spike beginning in Oct. 2016. However, when we look only at hits and take out the bots, it is clear that there is a consistent usage increase September to October with a peak in November for both 2015 and 2016. There is a slight dip in December but the lows are definitely in the summer months (June, July, August). This shows that activity by real people follows a pattern dictated by the academic calendar. The spike in visits that began in Oct. 2016 is likely a result of increased bot activity which has continued to the present, not because of increased use by real people.
To reiterate, Urchin may not have accurately filtered out all the bots. One way we decided to combat any possibility of misinformation was by looking at hits to the glossary page. There’s no reason for bots to be overly interested in the glossary page, but it would be something people would use. We decided that by focusing on the glossary we might get a better idea of how many real people use medart. I found that the glossary page appeared as one of medart’s top 5 most visited content for the past 18 months. The glossary had 9,074 visits and was the third most visited page. In comparison, the home page had the most visits at 89,012. The vast difference in these numbers indicates that a large portion of home page visits could be bots. 9,074 visits are much closer to what we would expect from 18 months of usage. For each time period, Urchin would only give me information on the top 5 pages in medart. I went through month by month to see if the glossary was always in the top 5 content. The glossary didn’t show up in the top 5 content at all until September, 2016. It is possible that the glossary rose in popularity as a result of the Sustaining Medart team’s visit to Kalamazoo in May, 2016.
Medart’s top 5 content consistently included the home page and the main menu pages for France and England, but interestingly enough, medart’s page on the Tower of London also appeared consistently. In fact, this page was the fifth most visited page overall for the past year and a half. The Tower of London page got 7,096 visits, and users spent an average of 6 minutes and 12 seconds on it. This is compared to the glossary which saw 3 minutes of average time spent. Of all the top 5 content, users spent the most time per visit with the Tower of London page. This activity did not sound like bot activity, so why were so many users spending so much time on this page? I dug a little deeper and found that the Tower of London was in the top 5 all 18 months except during July, August, and September of 2016 with a low in Dec. 2015 and a peak in March, 2016. This very nearly mimics the academic calendar, so it could be something used for classes. I wanted to know how people were finding this page so I googled “Tower of London Medieval Timeline” and looked under google images. An image from the Tower of London page was one of the first to come up. I found that someone had posted a link to this page on Pinterest. After some more searching I found two more images from medart saved to Pinterest. It’s exciting to see medart integrated into social media, and to see the effect it can have on site usage. It was good that I was able to get some insight on how actual people found medart. According to the Urchin Report, everyone who used medart got there by typing in the URL. This is extremely unlikely which is why I did some actual searching to get this information.
I did a similar search to find out how people were getting to the glossary. I googled “medieval glossary” and medart’s glossary was the ninth result. I checked through the other results to see if any linked to medart but didn’t find anything. People certainly can get to the glossary this way, but I think it’s more likely that usage increased because of the presentation at Kalamazoo in May.
Overall, the Urchin Reports presented us with a lot of really helpful information. We got further validation that medart usage revolves around the academic calendar, and we were able to look more closely at bot activity on the site. The Urchin Report also led me to discover other ways people have linked to medart, such as through Pinterest.
- Sustaining MedArt
- Graduate Work