RFM scores are built from three components: Recency, Frequency, and Monetary.
Though they are scored the same, they act differently.
Recency is an odd one and can look the most chaotic for some. That's because Recency will start to decay from the moment a purchase is made and get lower and lower until the customer buys again. Customers are always jumping to the top of it and pushing other customers lower down the ranking. In a high-volume store, that decay can be rapid.
For example, if John orders right now then he has the most recent order and would be at the top of the Recency. But when Sally orders a minute later, she's now at the top and John is in the 2nd place. Continue making sales and eventually John will drop to a lower rank (5 -> 4).
Frequency and Monetary decay too but only if the new order is for a customer who has ordered more often than John (Frequency) or spent more than John (Monetary). A brand new customer might not shift John's Frequency or Monetary down at all.
This automatic decay is useful for you as a store. It easily models how people's behavior change over time, e.g. the longer in-between orders the higher the chance of forgetting about your store. It also lets you adapt your marketing so customers shift into different campaigns with different goals (e.g. maximize revenue vs capture an order before it's lost).
That's why running your RFM analysis frequently is important and why it's best to have an app like Repeat Customer Insights run it for you. If you only look at them once a quarter or once a year, they will have shifted so much that you'll be starting over when it comes to your segments (in addition to having sent the wrong campaigns to the wrong customers).
Ideally you'd look at them once a week or maybe once a month if you don't have a lot of customers ordering each month.
Track down which customer cohorts perform the best
Different groups of people behave differently. Repeat Customer Insights creates cohort groups for you automatically to see how your customers change over time and spot new behavior trends.