Defining RFM thresholds on paper is hard. What counts as recent in retail is not the same as in fashion. Frequency changes with the purchase cycle. Monetary value often has long tails and a few spikes that distort everything. In this context, defaults rarely work well.
Cluster Assistant solves this. It reads your orders, analyzes how your data is distributed, and proposes the best cutoffs for the five levels of Recency, Frequency, and Monetary. You review them, refine them if you want, and apply them instantly.
This assistant, like the others in the Magnews Q family, requires acceptance of the service terms. Click the icon
to complete the activation process.
How it works in practice
Open your RFM model and start Cluster Assistant.
At the end of the data analysis phase, you will find two sections: Calculation and Statistics.
In Calculation you see the proposed thresholds for Recency, Frequency, and Monetary, shown next to the Previous values, with the star rating for levels from 1 to 5. You can edit every value before applying.
In Statistics you see the numbers that informed the calculation. It is the base for an informed decision.
Transparency, not extra work: the statistics
In the Statistics tab you find the numbers the assistant relied on. They are there to give you transparency, not extra work. In other words, you do not need them to set the thresholds, Q has already done it. If you want to understand how it reasoned, you have at your disposal:
- how many people were included in the calculation and the range of values (minimum, maximum);
- a balanced central measure (median) and the mean;
- variability indicators that resist spikes (such as the interquartile range and a deviation from the median);
- a signal of how the distribution leans (skewness).
These elements reassure you that the thresholds are not random. They come from the real shape of your data.
When it works best
Choose a time window that fits your purchase cycle. After unusual periods (sales, Black Friday, seasonal peaks) run a new calculation, the model stays aligned. After applying, check how contacts are distributed across levels. If a level is too full or almost empty, fine tune the cutoffs and reapply. It takes only a few minutes to reach balance.
What you need to get started
Order data with date, amount, and contact, already synced to magnews. An active RFM model to update. Magnews Q credits to run the calculation.
Quick questions
- Can I edit the proposed thresholds?
Yes, always. The proposals are a starting point, not a constraint. - What happens after applying?
The RFM model updates and the scores are recalculated at the next model run. Segments and automations that use RFM benefit immediately. - How long does the calculation take?
It depends on data volume. You can see the progress in real time.
In short: Cluster Assistant takes you from eyeballed thresholds to thresholds based on your data. Fewer attempts, more precision. And an RFM that truly helps you grow.