Resources
Churn prediction
How to spot churn before the customer is gone.
Recency, frequency, monetary: the classic customer segmentation, explained simply.
An RFM analysis ranks your customers by when they last bought, how often and how much. Vittorio, the sales assistant powered by artificial intelligence (AI), goes one step further and shows your team every morning which individual customer has an occasion today.
3 contact suggestions for today
RFM analysis is a proven method of customer segmentation. It ranks customers solely by their buying behaviour, without surveys and without details of age or place of residence. The name stands for the three values determined for each customer:
The idea behind it: whoever bought recently, often and a lot is closely attached to your business. Whoever used to buy often and a lot but has not been in for a long time is a candidate for personal contact.
Measured in days or months since the last purchase. In retail, this value is usually the strongest signal: the longer a customer is away, the harder it is to win them back.
The number of purchases or visits in the period under review. A clear rule on what counts as one purchase is important, for example all items on one day together.
The total revenue in the period. Calculating with gross margin instead of revenue rates high-margin customers more fairly.
The method is simple, and that is exactly why it is popular. For day-to-day work with regular customers, however, it has gaps:
If you want to see early on which customers are drifting away, continue with churn prediction.
Vittorio looks not only at segments but at each customer individually, with their purchases, their service and their occasions. The result is not a report but a suggestion for the responsible employee: whom to contact, why today and with which message. More on this on the AI customer data analysis page.
Example scenes. All names of people and businesses are fictitious.
Five customers, viewed over 24 months. Classes 1 to 5 result from comparison with all the store’s customers.
| Customer | Last purchase | Purchases | Revenue | Score | Segment |
|---|---|---|---|---|---|
| Customer A | 3 weeks ago | 6 | €8,400 | 5-5-5 | Top customer |
| Customer B | 2 months ago | 4 | €1,150 | 4-4-3 | Loyal customer |
| Customer C | 1 month ago | 1 | €2,300 | 5-1-4 | New customer |
| Customer D | 19 months ago | 5 | €11,200 | 1-4-5 | At-risk good customer |
| Customer E | 22 months ago | 1 | €180 | 1-1-1 | Dormant customer |
Illustrative figures. The potential analysis shows your own numbers, based on your data.
How to create a first RFM analysis with your own sales data.
For example the last 24 months. For each purchase, you need customer, date and amount.
Days since the last purchase, number of purchases and total revenue in the period.
Split each value into five equal groups. The best 20 per cent get a 5, the weakest a 1.
The three classes give the score, for example 5-5-5 for the best customers or 1-4-4 for good customers who have not been in for a long time.
Group similar scores into a few segments, such as top customers, loyal customers and at-risk good customers.
For each segment, decide who gets in touch and with what. Refresh the analysis regularly, for example every month.
Resources
How to spot churn before the customer is gone.
Resources
Six patterns in your data that become occasions.
Resources
What to do with at-risk good customers.
What your clients entrust to you stays in your house.
Every message your house sends on its own initiative is approved by a member of staff. Vittorio never writes on its own authority.
Contact exclusively with customers who have expressly agreed to WhatsApp.
Messages are sent via Meta’s WhatsApp Business Platform, not via private accounts.
Designed for use under the General Data Protection Regulation (GDPR), with a data processing agreement.
Before you decide on any software, we analyse your existing sales data, pseudonymised and contractually secured. After ten working days, we present the results to you in person.
Illustrative figures. The analysis shows your own numbers, based on your data.
Recency, frequency and monetary, meaning when a customer last bought, how often they buy and how much revenue they bring in.
For each customer, you determine the days since their last purchase, the number of purchases and the revenue within a period. Each value is divided into classes from 1 to 5. The three classes give the RFM score, from which segments can be derived.
It should cover several of your customers’ buying cycles. In retail, 12 to 24 months is common. For infrequent purchases, such as at a kitchen studio, it can be considerably longer.
Customer segmentation is the umbrella term for any grouping of customers. RFM analysis is one specific method for it that uses buying behaviour alone.
Yes. With a table of customer, purchase date and amount, the three values, the classes and the score can be calculated in a spreadsheet. It only becomes laborious when the analysis is to be refreshed regularly and turned into contacts.
It is a snapshot, does not take each customer’s own buying rhythm into account and knows no occasions such as service, contract end or anniversaries. It also treats all customers in a segment the same way.