Origin and purpose of RFM
Understand the three behavioural signals before touching the model.
Dynamic RFM is the behavioural intelligence component of a four-tool loyalty architecture. It must create useful analysis today without closing strategic decisions that belong to the Loyalty Strategy Planner.
Produces behavioural evidence, period movement and a customer-level SPO-ready RFM dataset.
Produces relationship intelligence at the same customer_key level.
Produces product/service intelligence at the same customer_key level.
The fourth tool merges RFM + NPS + ABCD using customer_key. Only there should the user inspect the SPO distribution, choose score ranges or percentiles, create named segments, assign economic objectives and loyalty mechanics, and calculate the final strategic plan.
This RFM tool does not calculate the final SPO and does not preselect Top 10%, Top 15%, Top 20%, Diamond or any other final loyalty segment.
The DOC ROI loyalty framework recognises three behavioural perspectives. They can coexist in the analytical engine, but the SPO-ready export must still contain exactly one row per customer_key.
R = recency of identifiable interaction. F = interaction frequency. M = attributable economic or commercial value.
R = recency of purchase. F = purchase frequency. M = monetary value.
R = recency of service/support interaction. F = service/request frequency. M = economic relationship impact or opportunity.
If several RFM layers are available, the tool places them in different columns of the same customer row. It never creates three SPO-ready rows for the same person.
Post-sale activity is behavioural evidence, not satisfaction. NPS provides relationship sentiment. If you already have a validated Content or Post-sale RFM score, map that optional score directly; otherwise this tool derives a relative 1–5 behavioural score from the mapped metrics.
Dynamic RFM compares two equivalent periods. The export uses only three movement values: Growing, Stable and Declining.
Growing
The current behavioural band is stronger than the previous period.
Stable
The current behavioural band remains at the same level.
Declining
The current behavioural band is weaker than the previous period.
The tool combines the available RFM evidence with period movement to generate cognitive_score as an integer from 1 to 5, cognitive_category as 1-star to 5-star, and cognitive_movement using the exact movement terms above. This is a behavioural interpretation, not the final SPO.
The scoring configuration is editable. The customer population is not. A real source with N unique integration keys must return N SPO-ready customer rows.
1. Comparable periods
P1 and P2 have the same duration. The SPO-ready fields period_p1 and period_p2 store the corresponding period-end dates in ISO YYYY-MM-DD format.
2. Transaction R/F/M weights
Weights are normalised automatically. These weights create the analytical transaction RFM score, not the final SPO.
3. Transaction factor ranges
Classroom defaults for a 90-day Don Espadín exercise. Edit the thresholds to match your own purchasing cycle and economics.
| Factor | Recency from | Recency to | Orders from | Orders to | Revenue from | Revenue to |
|---|
For real data, select the existing field that uniquely identifies the customer: CRM customer ID, loyalty member ID, hashed email ID, ERP customer ID or another stable identifier. The tool preserves it as customer_key and also preserves the original local identifier as source_customer_id. Row number is never used as an integration key.
1. Dataset and source
Use the fixed Don Espadín demo or upload your own CSV. The demo contains exactly 266 fictional customers: DE-0001 through DE-0266.
2. Review / paste the source table
Transaction fields are required for the Dynamic RFM period engine. Content and Post-sale fields are optional customer-level layers.
3. Map the integration key and behavioural fields
The selectors below are populated from your source headers. Optional fields can remain “Not available”.
Mandatory identity
Transaction RFM · required for period comparison
Content / Marketing RFM · optional
Post-sale / Relationship RFM · optional
recommended_action, recommended_channel and estimated_cost are operational RFM recommendations. The final economic objectives, loyalty mechanics and named strategic segments belong to the Loyalty Strategy Planner / SPO Builder.
Calculate behavioural intelligence
The calculation preserves every unique customer key. The final SPO-ready output must contain the same number of rows as unique customers in the source.
Behavioural Portfolio Intelligence
Don Espadín classroom case or production portfolio · one customer_key per customer
Cognitive Category distribution
Current behavioural interpretation from 1-star to 5-star. This is not the final SPO.
Recency × Frequency risk map · P2
Red cells identify weak intersections of R and F. Counts show customers in each P2 behavioural cell.
RFM band migration · P1 → P2
Rows = P1 analytical band. Columns = P2 analytical band. Downward movements are highlighted in red.
Average ARPU by Cognitive Category
ARPU is calculated at category level from P2 transaction revenue and written as a numeric value into each customer row.
Behavioural layer availability
See how many customers contribute Transaction, Content and Post-sale evidence.
The export is an interoperability product, not merely a classroom download. Its structure never changes with portfolio size: N unique customers in = N customer-level rows out.
30 required columns: schema_version · dataset_id · customer_key · source_customer_id · period_p1 · period_p2 · content_rfm_available · content_recency · content_frequency · content_value · content_rfm_score · transaction_rfm_available · transaction_recency · transaction_frequency · transaction_monetary · transaction_rfm_score · post_sale_rfm_available · post_sale_recency · post_sale_frequency · post_sale_economic_impact · post_sale_rfm_score · cognitive_score · cognitive_category · cognitive_movement · segment_p1 · segment_p2 · recommended_action · recommended_channel · estimated_cost · arpu
Validation checks key uniqueness, cardinality, cognitive score, movement values, numeric cost/ARPU, ISO dates and the absence of currency symbols in numeric fields.
Download message: “This file contains one standardised row per customer and is ready to merge with DOC ROI NPS and ABCD Product Intelligence inside the Loyalty Strategy Planner.”
25 customer rows are visible at once. Scroll inside the table to inspect the generated customer population.
Dynamic RFM supplies behavioural intelligence. NPS supplies relationship intelligence. ABCD supplies product intelligence. The Loyalty Strategy Planner merges the complete customer population by customer_key and builds the final SPO strategy.