The dashboard says conversion fell.
Does that justify changing onboarding?
Harbor Market · acquisition export
Authored practice data| Channel | Week 1 users | Week 1 converters | Week 2 users | Week 2 converters |
|---|---|---|---|---|
| Direct | 900 | 180 | 100 | 24 |
| Paid | 100 | 10 | 900 | 108 |
Metric contract and missing evidence
One row represents one eligible new user. Channels are mutually exclusive and assigned at first acquisition. A conversion is at least one completed purchase within seven days. Both windows are closed. Eligibility and event capture are unchanged.
Paid acquisition expanded in week 2. Spend, margin, random assignment and customer characteristics within each channel are absent.
How the comparison works
For each channel, divide converting users by eligible users. The total rate weights those channel rates by their user counts. Holding the earlier mix fixed separates the arithmetic effect of rate changes from the effect of a different mix.
This decomposition describes the data. It does not identify the cause of a change or show whether paid acquisition adds profitable customers.
Airbnb: observed versus incremental value →