RFM segment sizes can change as user behavior, the eligible population, and the data included in RFM calculations change over time. Use this guide to understand why RFM segment sizes change, investigate unexpected changes, and determine which historical data you can review.
Why can RFM segment sizes change?
The number of users in an RFM segment can change for the following reasons:
Recency changes: A user’s Recency score can decrease as more time passes since their last purchase or activity.
Rolling data span: RFM processes up to 12 months of data. Older activity stops contributing when it falls outside the calculation span.
Relative scoring: RFM scores rank users against the current eligible population. They are not fixed spending, purchase-count, or recency thresholds.
Daily reclassification: The daily calculation can move users from one named RFM segment to another.
User-profile retention: A profile that reaches the applicable User TTL is removed. The user can then no longer belong to an RFM segment.
A decrease in one RFM segment does not always mean that users were deleted or that data was lost. Some users might have moved to other RFM segments.
For example, a user has frequent, high-value purchases but does not purchase again for several months. Their Frequency and Monetary scores might remain high while their Recency score decreases. The daily RFM recalculation can then move the user from Champions to another RFM segment.
A larger audience change can occur when several factors overlap. For example, some users might receive lower Recency scores while other profiles reach the User TTL.
How can I troubleshoot an unexpected RFM segment change?
Use numbered steps because this is an investigation flow.
Confirm the timing. Check whether the change appeared after the daily RFM update.
Review recent activity trends. Check for changes in purchase volume, frequency, revenue, or the activity used by the enabled RFM model.
Review the calculation span. Compare the affected period with approximately 12 months earlier to identify activity that might have left the RFM data span.
Compare related RFM segments. Check whether users moved from the affected segment into another RFM segment.
Consider User TTL removal. Check whether a group of user profiles might have reached the applicable User TTL.
Review Segment Analytics. If tracking was enabled before the change, use the daily snapshots to identify when the segment started growing or shrinking.
How can I compare related RFM segments?
Do not review the affected RFM segment in isolation. A decrease in Champions or Loyal Customers might correspond to an increase in another RFM segment.
Compare related segments across the same dates to determine whether:
Users moved between segments.
Several RFM segments decreased at the same time.
The total eligible RFM population also decreased.
A decrease in one segment with a corresponding increase in another suggests reclassification. A decrease across several segments and the total eligible population can point to a broader population change, such as profile removal.
What historical data can you review?
Question | What you can review |
|---|---|
What is the segment’s current size? | The current RFM segment count |
When did a tracked segment grow or shrink? | Daily snapshots in Segment Analytics |
What was the segment size before tracking started? | Not available through Segment Analytics |
Which users entered or left on a past date? | Not available as a complete retrospective membership timeline |
What previous RFM scores did a deleted profile have? | Cannot be reconstructed from the deleted profile |
Why did one user’s score change on a historical date? | Can be investigated only if the required RFM attribute and calculation history are available |