New vs Returning Users
New vs Returning (NewVsOld) splits users into new/returning groups and compares their differences in scale, behavior, and monetization, answering "how do new users differ from returning users". It corresponds to the backend AnalyticsService.NewVsOld.
1. What Questions It Answers
- What share of current actives are new users? How many are returning?
- How big is the gap between the new-user payment rate and the returning-user payment rate?
- Are new users' event volume (activity level) lower than returning users'? (an early-churn signal)
2. Comparison Dimensions
Each NewVsOldSegment (new/returning) contains:
| Metric | Meaning |
|---|---|
user_count | Number of users in the group |
event_count | Total events in the group |
pay_users | Number of paying users in the group |
pay_rate | Payment rate of the group (0-1) |
User types: new (new user) / old (returning user). A user is "new" if registered within the last new_user_days (default 7 days).
3. Backend API
gRPC: AnalyticsService.NewVsOld
| Field | Type | Description |
|---|---|---|
time_range | TimeRange | Analysis time range |
app_id | uint32 (optional) | Filter by app |
new_user_days | uint32 (optional) | Days defining a new user (registered within the last N days), default 7 |
Response NewVsOldResponse: segments[] (two NewVsOldSegment: new / old).
HTTP (admin proxy)
POST /admin/v1/analytics/new-vs-old
Content-Type: application/json
{
"timeRange": { "startMs": 1718169600000, "endMs": 1718774399000 },
"newUserDays": 7
}
4. Typical Scenarios
Early-stage payment conversion
Compare new vs returning pay_rate: if new users pay far less than returning ones, early-stage payment conversion is the bottleneck — improve onboarding or offer newbie discounts.
Activity level differences
Compare event_count / user_count (events per user): low per-user event counts among new users may mean onboarding failed to engage them — a high churn risk.
Balancing acquisition vs retention
Look at the user_count share of new/returning: a new-user share that is too high (rapid influx) may dilute overall quality; a returning-user share that is too low signals poor retention.
5. Notes and Caveats
- The new-user threshold is adjustable:
new_user_daysdefaults to 7 days; longer-cycle products can raise it (e.g. 30 days), shorter-cycle ones lower it. - Depends on registration time: the new-user definition is based on
register_time, which requires registration events to be collected correctly. - Binary comparison: it only splits new/returning, with no finer tiers (e.g. "recently new"); for finer stages use User Lifecycle.
- Payment rate basis:
pay_rate= paying users / users in the group, affected by whether payment events within the time range were reported.