List performance with overlapping lists – accurate?

[MMiller18]

We would like to do RFM analysis against our HubSpot marketing emails.

My question is: how can we get accurate results using RFM analysis against HubSpot lists that are overlapping (some recipients appear in both lists).

By definition,  lists based on RFM are highly likely to overlap.  Eg a High Frequency list (frequent buyers) would almost certainly also contain recipients in a list based on Recency (recent buyers).

• I know that HubSpot can show performance of a given email by list.  But wouldn’t the results be skewed by the fact that the lists are overlapping?

[karstenkoehler · Accepted solution]

Hi @MMiller18,

I'm not familiar with RFM but it seems like the methodology generally allows for overlapping segments, at least according to generative AI:

---

Overlapping Segments

1. Acceptance of Overlaps:

   • Overlapping segments are generally accepted in RFM analysis. For example, a customer can be classified as both high frequency and high monetary, indicating they purchase often and spend a lot.

2. Prioritization:

   • Some businesses may choose to prioritize one dimension over others based on their specific goals. For example, if customer retention is a primary goal, recency might be prioritized. If increasing revenue is more important, monetary might take precedence.

Common Practices

1. Composite RFM Scores:

   • Combine RFM scores into a single composite score to create distinct segments. For example, a customer with scores of R=5, F=4, M=5 might be labeled as a "high-value customer," while a customer with R=1, F=1, M=1 might be labeled as a "low-value customer."

2. Separate Analysis:

   • Perform separate analyses for each dimension to understand how customers behave in terms of recency, frequency, and monetary value. This can help tailor specific marketing strategies for different segments.

3. Segment Overlap Analysis:

   • Recognize and analyze the overlaps to understand multi-dimensional customer behavior. For example, a Venn diagram can be used to visualize the overlaps between high-recency, high-frequency, and high-monetary segments.

---

It seems to me like you could analyze each RFM list and draw conclusions from it – being aware that one segment does indeed overlap with others. As long as this is considered in reading the findings, you can still generate insights from that. As long as the % of overlap is part of an annotation and added context, I (as a non FRM expert) would find this acceptable.

If you want to have lists without overlap, you would have to prioritize the lists, rank them and then make sure that list 2 excludes contacts from list 1, list 3 excludes contacts from 1 and 2 etc. – however, this seems to skew things more than accepting overlaps.

Hope this helps!