Goals:

• If a user doesn't do their scheduled burst activities after N days, we send them an SMS notification.

Non-Goals:

• Generic re-usable automated notifications.

How to Model Activity Burst

Requirements

mPower 2.0 is a new study, so we don't have to deal with users upgrading from mPower 1.0, so we can use enrollment to trigger study bursts.

A user will have a study burst every 13 weeks, which roughly divides the year into 4 quarters, and starts and ends the study burst on the same day of week every burst. A study burst will last 2 weeks. There will be 9 study bursts, lasting 2 years + 1 extra study burst at the end of the two years.

Implementation

When a user is enrolled in the study, Bridge Server will automatically generate custom events activityBurst1Start, activityBurst2Start, ... , activityBurst9Start. (We can't have duplicate event IDs, so each event ID needs to be different.)

In Bridge Server, we'll have a schedule set up, scheduled off of each of activityBurstNStart, which schedules daily events for the next 2 weeks. (Note that for custom events, we add the prefix "custom:" to the event key. Example: "custom:activityBurst1Start".)

When the user first installs the app, the app will ask Bridge for events for the next 2.5 years. This will generate all the events on Bridge Server. This means we can assume the events will always exist for each user. (IMPORTANT: Bridge Server currently limits the Get Activities call to a max of 15 days. In order to get activities for the next 2.5, the app will need to make Get Activities calls 15 days at a time.)

When the app completes each activity, it will need to call the Update Activity API to mark that activity as completed.

New Feature: Automatic custom events per study.

New Feature: Bridge can schedule off of each of a list of events, rather than one of a list of events.

Alternatives

Alternative #1: Create 9 different schedules, one for each study burst. This means less work on Bridge Server, but it requires us to manually create 9 identical schedules. This is not recommended, as maintaining 9 identical schedules leads to replication errors.

Alternative #2: Instead of creating a bunch of custom events, we should implement compound schedules, eg daily activities every 2 weeks, repeat the schedule every 13 weeks for 9 iterations. This is not recommended as it increases the complexity of the Bridge Scheduler tremendously. It would be much simpler to simply create two very simple features (automatic custom events, and schedule for each event) and combine them than to make one very complex feature.

Sending Notifications

We want to be able to send SMS notifications on an individual basis. Topics and subscriptions doesn't really fit the use case, so we'll need to create a worker facing "send SMS to user" API. Note that this message should be marked as "Promotional" rather than "Transactional", since it's not really transactional in nature.

SMS vs In-App Notifications

SMS notifications are desired specifically because they are more heavyweight than in-app notifications, and because users who haven't engaged with the app are unlikely to respond to in-app notifications.

Opting Out of SMS Notifications

We need to give users a way to opt-out of receiving SMS notifications. AWS handles this automatically for us (see https://docs.aws.amazon.com/sns/latest/dg/sms_manage.html#sms_manage_optout). Part of the notification should inform the user that they can choose to quit the study replying with QUIT.

As a supplement, we can have an SMS Notification setting in the app, which can mark the user with a data group (or profile attribute). The Notification Worker can then filter on these data groups (or profile attributes).

Important: SMS opt-out is account wide. If a user opts out of SMS from Bridge, this means they also no longer receive SMS to verify their phone or get a phone sign-in link for any study. Note that email opt-outs are only 0.2% of our emails. We expect SMS opt-out rates to be lower. Therefore, this is unlikely to affect more than a small handful of users. We can look into SMS legal requirements and granular opt-outs as a Stretch Goal, but we've decided that it's not an mPower 2.0 launch blocker, nor is it necessary to design this out upfront. Also note that, unlike email, we don't get any notifications if a user opts out of SMS. We have to manually poll SNS to determine users who have opted out. As a Stretch Goal, we can build a system to poll users who have opted out of SNS and do any necessary follow-up.

NOTE: It was determined that a user opting out of SMS notifications isn't required to leave the study.

In order to prevent spurious SMS sign-in messages, we want to make Reauth more robust. See BRIDGE-2154 - Getting issue details... STATUS

Activity Burst Notification Worker

We will need to add worker APIs for:

• get activity events for user
• get activity history for user
• send notification to user
• (optional) get schedule plan

(NOTE: These APIs all exist for study researchers, but they do not yet exist for workers.)

The worker will be scheduled to run daily at (for example) 20:00 UTC (12pm PST / 1pm PDT). The worker will iterate over every user in the given study, and for each user:

1. Check the user's data groups and/or profile attributes to see if they are eligible for SMS notifications. If not, skip.
2. Look up the user's activity events to determine where they are in the study burst.
3. If the user is in an activity burst, get all their activities since the start of the activity burst to see if they've been completing their activities. We can either check all of a user's activities, or we can check the user's activities for a specific schedule.
4. If certain conditions happen (the user has done less than X activities in at least Y days), send a notification to the user using the Send Notification API.

Handling Time Zones

(Stretch Goal)

The previous design assumes everyone is in the US, and it will send notifications at 12pm/1pm for PST/PDT, 1pm/2pm for MST/MDT, 2pm/3pm for CST/CDT, and 3pm/4pm for EST/EDT. Note that Alaska is one hour before Pacific Time, and Hawaii is 2 hours before Pacific Time, meaning that Hawaiians can be notified as early as 10am. This doesn't account for US residents in Pacific Islands that are far outside our timezones, or are otherwise not anywhere near contiguous US timezones. This also means that the notification can be spread around up to 4 different hours in the contiguous US (more outside the US), rather than having a consistent notification time.

We can instead have the notification worker run every hour, calculate what timezone offset corresponds to "noon local time" (or whatever arbitrary local time we want notifications to be sent at). Note that special handling will need to be done to handle fractional timezones. (Note that there are no fractional timezones in the US.) This will be configured as part of the notification worker.

While iterating users, the notification worker can check the user's time zone (this is persisted when they first ask for scheduled activities), and if that user's time zone doesn't match with the time zone we're currently processing, we skip the user.

Note that there are a few issues with this approach:

• Note the users cannot currently change their timezone on the server. Even during DST change, we use the user's original timezone for scheduling. This means that during DST change, or if the user physically changes timezones, this could cause the notification to be sent at a different local time zone.
• If in the future, we do allow a timezone change, this may cause the notification worker to "skip" that user. This can be fixed by having our notification rules to send on Day >= 3 instead of on Day 3, but we'll need to keep track of whether we've sent the notification to that user for this activity burst, to make sure they don't receive duplicate notifications.
• The notification worker iterates over all users 24 times a day. This is fine for now, while the participant pool is fairly small. Long term, we could solve this by building a more sophisticated worker, where the first worker runs daily, goes through all users, figures out if we should send a notification, and if so, figures out when (based on user's time zone). The second worker runs hourly, figures out which users have scheduled notifications, and then sends them.