Apostello is an open source SMS platform that was built for St Columba’s Free Church. It has been in production since January 2015. For more details on apostello, visit the Github repo or read the documentation.

Since apostello has been in production for over a year, I thought this would be a good time to look at the usage patterns. All the graphs were generated using a Jupyter notebook: source, in case you want to repeat this analysis on your own apostello data.

Inbound Messages

Keyword Histogram

In total 384 messages have been sent to apostello, that’s an average rate of 1.04 per day.

The histogram above splits the messages according to the keyword they matched. We can see that:

  • “connect” (143) is by far the most popular keyword
  • Unfortunately, “None” is the next most common, about 17% of the incoming messages did not match a keyword. Considering the system is setup to always expect a keyword, we may need to better communicate that with the end users
  • Only one opt out (“stop” keyword) message was received

Time Distribution

The above plots have segmented the messages into the time of day, the day of the week and the month of the year the messages were recieved. You can see:

  • Fairly even usage distribution throughout the day (and surprisingly there is no clumping around the “connect” deadline at 11pm)
  • Most messages are received on a Sunday, followed by a Wednesday. The Wednesday bump is probably a result of notices in St Columba’s weekly email
  • There was a clear lull in activity over the summer months. This will be because “connect” does not run during the summer

Outbound Messages

Time Distribution

  • We have sent a total of 943 messages - about 2.5 per day
  • For every message we have received, we have sent about 2.5
  • Most messages were sent in the morning and on Sundays. This is because apostello was used to alert the congregation to a couple of last minute changes to service times caused by unforseen circumstances, artificially inflating this figure

Summary

  • Total throughput of 1327 messages
  • “connect” is easily the most popular keyword
  • We seem to send many more messages than we receive
  • More communication about keywords is required