day_1_return
Shows whether the user remembered the site after first visit.
Do not send aggressive reminders to grow D1.
return without dependency
Retention
How to measure returns to My Day, journal, inbox, favorites, and account without creating an anxious habit.
Good retention makes the site a useful ritual, not an anxiety button.
journal_entry_created
Shows that the user is building a personal sign history.
The journal must not ask for medical or legal data.
inbox_message_opened
Shows usefulness of Oracle letters.
Low open may be a subject-line or deliverability issue.
Collect the signal
Define the event, page, source, and moment where the user acted or stopped.
- The event does not contain a private question or full scroll text.
- There is a clear timestamp and route.
- The signal can be tied to an owner decision.
Compare against expectation
Every number must be read beside a hypothesis: what should have happened, what happened, and how much it matters.
- There is a baseline or first-week manual estimate.
- No conclusion is made from one random day.
- Devices are checked separately: mobile and desktop.
Turn into an edit
Analytics becomes useful only after action: rewrite a block, simplify a flow, add FAQ, fix an error, or close an extra door.
- There is an owner, priority, and expected effect.
- The change is written into the decision log.
- A re-check date is assigned after the change.
dashboard
- D1, D7, D30 returning users by entry page.
- My Day usage and journal saves.
- Inbox opens by letter type.
- Repeat purchase spacing and theme diversity.
alerts
- Repeat purchases happen too frequently around the same theme.
- D7 drops after a My Day change.
- Inbox opens but does not lead to a useful action.
decisions
- Add gentle pause copy if the user repeats the same question.
- Improve journal prompts if returns exist but entries are not created.
- Rewrite inbox categories if letters are not opened.
red flags
- Retention is built on fear of missing an answer.
- The site pushes another reading instead of pause and observation.
- The journal collects unnecessary personal data.
related doors