Page 6 of 7 — Finding the threshold that separates retained readers from churned ones.
Weekly retention by number of people followed in first 7 days.
Each line shows what percentage of users in that cohort were still reading each week. The separation is dramatic: users who follow ≥5 people in their first week maintain ~45% activity at month 3, while those below decline to 13%. The curves above ≥5 continue to improve but with diminishing returns and rapidly shrinking populations.
Following thresholds compared at weeks 4, 8, and 12.
| Threshold | Users | Week 4 | Week 8 | Week 12 | W8→W12 | Gap at W12 | Penetration |
|---|
The key column is W8→W12 — how much the retention curve declines (or stabilizes) between months 2 and 3. At ≥1 and ≥3, the curve is still declining fast (−8pp and −7pp). At ≥5, the decline halves to −3pp — the curve is flattening. At ≥7 and above, it’s nearly flat. This is the Facebook-style finding: ≥5 is where the retention curve transitions from “still losing users” to “stable retention.”
The activation curve that defines Bookwise retention.
This is the Bookwise activation curve. Users who follow 5 or more people in their first week are 3.5× more likely to still be reading at month 3. The gap opens immediately in week 1 and remains stable through month 4.
198 users (43% of the dataset) signed up during December 2025, likely tied to the Dragonsteel Nexus event. Do they behave differently?
The magic number holds across both cohorts — users who follow ≥5 people consistently outperform those who don’t. But the effect is dramatically stronger for organic users (39pp gap vs 16pp). The December cohort’s lower overall retention suggests event-driven signups with lower intrinsic motivation. For experiment targeting, this means: the magic number is most powerful for users who came to Bookwise with genuine reading intent, not event-driven curiosity.
Comparing the best threshold for each social metric.
| Metric | Best Threshold | W12 Retention (met) | W12 Retention (not) | Gap | Penetration |
|---|
Following has the largest retention gap of any actionable metric. Kudos received is close (29pp) but is downstream of reading — you can’t get kudos without generating content. Followers are a weaker signal because they reflect others’ behavior, not the user’s own agency. Following is the metric we can most directly influence through experiments.
195 mid-range readers — users who showed reading intent in their first week but didn’t form enough social connections — are the prime targets for our hackathon experiments. If we can move even 20% of them past the ≥5 threshold, we’d expect roughly 39 additional retained readers at month 3.