Anonymous usage statistics

[gmaclennan]

We currently do not track any metrics about Mapeo Mobile users, which means we do not have a clear idea about how many users we have.

We do have statistics about the number of active users who download Mapeo from the Google Play Store, but we do not have reliable data for users who download the .apk file and install via "side-loading" the .apk onto the device. We do have stats from Github about the number of downloads from the releases page, but we do not currently have data on downloads from https://apk.mapeo.app (which has only recently been added as a link on our website and does not currently work). Downloads are not a great indicator of installs, because for trainings a single .apk download can be used to install multiple phones. Today (May 27th 2020) I added logging to our AWS S3 buckets to track downloads from there.

It would be useful to have a better indicator of the number of active Mapeo users.

One option is to send anonymous usage data (similar to what is already collected by Google Play) the records active users (e.g. pings a log server whenever a user has opened the app). Ideally this would record usage events offline and then send data once a phone gets online. It still would not capture a large number of users whose phone is never online.

Another option is to use data from the mapeo-core database about the number of users within a project, and anonymize that data and send it from any device within a project that does get online. That feels more invasive to me, but we do not need to know any data about users other than whether they are using Mapeo so if we anonymize it all maybe it could be ok?

For both options we need to decide if it should be opt-in or opt-out. For the first option it would be per-device, where-as the second option would probably be a per-project option. We should probably expose the option prominently when somebody starts using Mapeo, and also have it as an option in the settings screen. I would prefer opt-out (e.g. collect anonymous usage stats by default) since this data is not very sensitive and is less than probably most other apps the user has installed (Android + Google is likely already tracking much more than that). But I think we should have a clear screen that the user sees before they use Mapeo that gives them a clear option to opt-out.

I am not sure if there are existing services for tracking users. It would be good to combine data from Google Play, bug reporting, downloads and active users, and that is a lot of work to set up a custom solution just for ourselves that has a way of accessing the data that does not require technical knowledge (e.g. so that grant writers can access the latest data easily).

[gmaclennan]

I did a quick review of services we could use for this (since setting up our own infrastructure for analytics is costly in terms of time and resources):

• Segment
• Mixpanel
• Amplitude
• Firebase Analytics

Firebase Analytics is free, but it's owned by Google and free comes at a price of data sharing.

Based on price, Amplitude seems the best option, because we would stay well within their free tier.

Needs further research and investigation, particularly around privacy policies and how complex the integration is and usability of the dashboard for viewing analytic data (we don't want something too complicated and overkill for our needs)