We have all become ‘zoomers’
In mid March, most universities announced a transition into ‘remote learning’. For nearly all schools, this means learning to love web conferencing tools. At many organizations, the communication tool of choice curing the COVID-19 quarantine is Zoom, a video conference tool experiencing explosive user growth over the past months.
Let’s take a moment to explore how this previously unknown tool Zoom has been adopted at a medium size, private university in the United States. Remote learning officially began at this university following the end of spring break on March 23 and the data extends 6 weeks to May 3.
Disclaimer
While the core feature set of Zoom is consistent across most organizations, the specific features available and their adoption vary considerably by organization. This work is only a single example of adoption in higher education and is not necessarily generalizable to other organizations or even other universities. It is presented as an interesting anectdote and not the basis of any proposed action.
Minutes Spent on Zoom
One of the metrics provided by the Zoom API for measuring platform utilization is the ‘participant minute’. A participant minute is defined as a single conference call participant spending one minute on a zoom call hosted by a university enterprise account. For context, a typical undergraduate course at this university has 20 students and requires 150 minutes of lecture per week. If we assume these values are perfectly replicated in the remote learning environment, we would expect to see 21 total participants (20 students and 1 professor) spending 150 minutes total together. This would equate to \(150 * 21 = 3150\) participant minutes per week for each undergraduate class.
While this is a useful example for explaining a participant minute, it is an unreasonably high expectation of use. Zoom is not required for lecture delivery and some classes are held asychronously with content delivered via online videos on platforms like YouTube. Also, roommates are choosing to participate in a lecture together, using only one participant ID.
Lastly, course lectures are far from the only use of Zoom on campus. Office hours, university committee meetings and college operations are all carried out using Zoom. As we will shortly see, it appears non-university social activities are also held using the university provided service.
Lets start by looking at how the total number of participant minutes spent on Zoom has evolved over the 6 week period.
The records for highest single day use occured early in the remote learning experience. The highest daily value peaks at just over 550,000 total participant minutes. This is about 7% of the more than 8 million total minutes logged. All of the three highest use days coincide with university and college orientation meetings for all faculty, staff and students. The first Monday following spring break is especially interesting because many of the calls occuring on this day had only a single participant. We can only assume faculty and students were anxious to try Zoom for the first time and wanted to properly configure their devices.
Platform use is largely consistent across the 6 weeks, with slightly lower use in the first three weeks compared to the following three. Most of the curiosities in this data appear to coincide with the specific day of the week, so that is where we will focus next.
Eh, I don’t want to work on Friday
If you are not familiar with university scheduling, faculty and students generally prefer to get in-person classwork finished before Friday. The most desirable class schedule meets twice a week on Tuesdays and Thursdays. The runner-up is a similar Monday and Wednesday schedule. Classes held on Friday are so unloved by students and faculty alike, they almost always have fewer students enrolled in them compared to classes on other days of the week. The university even opts to leave many building resources go unused on Friday rather than schedule unpopular courses.
This preference is alluded to in the previous visualization, but we will dig a little deeper by looking at the average number of Zoom calls held on each day of the week.
With over 40,000 calls held across the 6 week period, it is no surprise relatively few of those calls occur over the weekend. With that said, Saturday has nearly 2/3 the volume as Friday. The university does have Saturday courses, however most of them are graduate classes. This may suggest a high graduate utilization over the weekend or something compeltely unrelated.
As expected, there is clearly a preference for not Zooming on Fridays; however, Thursdays also have an appreciably lower number of calls compared to the rest of the work week. In light of the preferred Tuesday and Thursday meeting schedule during a normal semester, this is somewhat curious. Perhaps classes held on Tuesday and Thursdays have decided to front load their Zoom time on Tuesday and skip Thursday sessions all together. Of course, university politics also dictate more senior faculty are given the most desireable schedules and they may teach different courses than their less senior counterparts. Regardless, we lack enough information to conclude why this is the case.
What we fail to see in this visual is how the number of participant minutes correspond to the day of the week. Lets explore this by investigating both the total number of participant minutes spent on each day and the average number of participant minutes spent on each call.
The most interesting result here is how tightly spread the total daily participant minutes are for Thursdays. While there are fewer calls on Thursday than earlier in the week, it is still much higher than the results of Friday, Saturday and Sunday. The lower outlier on both Thursday and Friday is not surprisingly the week before Easter where the university has an extended break. Curiously, the Saturday before Easter is the highest volume day Saturday in the data and the Monday following Easter is likewise a high volume day despite being a holiday. While some of the Saturday behavior we may be able to explain momentarily, the Monday behavior is genuinely puzzling.
For a little more insight, lets look at the average number of minutes per call across each day of the week.
Accounting for the number of calls on each day narrows the spread a bit on Monday and Tuesday but Thursdays are still appreciably narrow and Wednesday’s average is still decidedly wide. The best explanation we have for the tight Thursday spread is some form of regularly scheduled calls with a consistent number of participants. Wednesdays are likely the day where most irregular, non-course-related meetings occur such as college and committee meetings.
Lastly, the average number of participant minutes per call widen on Sunday, even though the median is quite low at just over 100. Saturday on the other hand more than doubles this value and even exceeds the average and median values for all but Tuesday. If you recall from an earlier visual, Saturday also has nearly two-thirds the call volume as Friday, so this result cannot be blamed on a low number of calls alone. What gives?
Who is participating in these calls?
Every call logged in the data is hosted by a university employee using a university email address. On the contrary, participants are not even required to have a university email to join a call, let alone use it as their method of joining. Some calls optionally do not require an email address at all, instead being facilitated thorough traditional telephone services. For this final analysis, lets use this information to answer the following question: “What percent of daily participant minutes are logged by university accounts compared to non-university accounts?”
While most members of the university community use their university provided account to connect to Zoom, it is of course possible for a member of the university community to join a call using a personal email address. This means the following analysis cannot differentiate between non-university participants and university participants using a personal email account. Even in spite of this limitation, the results are striking:
During the work week, approximately 70% of all participant minutes are enjoyed by individuals using their university email to connect. Given the tight consistency across every weekday, it is reasonable to assume most of the remaining 30% are members of the university community using personal email accounts. A minority of calls are also used for connecting to non-university participants such as members of the local community and various committees and boards the university interacts with.
The weekend transition is hard to ignore. Saturday may be partially explained as having a higher proportion of graduate students and working professionals taking weekend degree courses and using their personal or work email accounts. This is a less defensible position in light of the activity observed on Sunday, where over two-thirds of all participant minutes are recorded by non-university email addresses.
Face it… we are all using our work Zoom accounts to have weekend dinners and Zoom parties with our friends also wallowing in quarantine.