A beginners guide to Recency Frequency Monetary analysis for B2B marketers

RFM analysis, which stands for Recency, Frequency, and Monetary value, is a marketing technique that quantifies and evaluates historic customer value. Why should B2B marketers bother with RFM? Because it offers a way to prioritise the use of marketing and sales resources and thereby maximise return on effort invested. By identifying and cutting out the activities that deliver low returns, you can focus on what really matters: raising response rates, improving conversions, and increasing the value of the average sale.

Where does RFM come from?

RFM originated with direct mail marketing organisations and their realisation that customers are not created equal. In fact, the typical distribution of customers by revenue for any given period of time is unlikely to follow a bell curve - in practice, a chart of customer value is a power curve with a long tail (Pareto distribution). Why? First, because customers have different spending power. Second, their actual propensity to buy varies enormously as well.

How does RFM work?

A brief look at the direct mail business model explains the basic questions that RFM was invented to answer. In classic direct marketing, the production and distribution of a new catalogue represents high up-front investment costs. Given that fact, direct marketers have to decide who gets a copy of the new catalogue and who doesn’t. How? They calculate the total costs of the catalogue and divide by the print run to get the average unit cost. Then add a margin for profit. Any customer whose sales revenue is less than that value is unprofitable, so they don’t send a new catalogue.

Next question: how to use the promotional budget? Answer: for a given category of goods, send the promotion to the people who are likely to spend more that the cost of the promotion. Repeated analyses over time showed marketers that three characteristics come into play. The people most likely to respond to a promotion are those who ordered most recently. Loyal customers who buy frequently (even though in small quantities) are a good bet, as well. And so are the big spenders (even though they might be infrequent buyers).

Hence RFM analysis. Each of the three characteristics – Recency, Frequency and Monetary – follows a power curve. So, if you create a composite based on those characteristics, you’ll get a list of customers, accurately prioritised by their historic propensity to buy. You then use this as a way to create target segments in the database.

Return on Investment

What’s changed since the days of direct mail catalogues? Well, we use websites and eShops instead of print catalogues. Promotions are sent via email instead of postal mail. The change in technology has cut the time-lag between triggering the promotion and receiving 80% of responses, from between 10 and 12 days to 24 hours.

But the key factor has not changed at all - resources. The most important fact – and yet it’s the one that B2B marketers are most likely to overlook – is that staff effort should not be treated as a sunk cost. If you’re serious about getting ROI out of direct marketing, then staff effort should be considered as opportunity cost. The key question is: ?given that we can only use this working day once, what offer is most relevant for a specific target group and will trigger the best response?“ RFM will give you an answer.

Tailored Marketing Campaigns

For marketers, grouping customers into segments based on RFM scores allows you to focus on the specific needs and preferences. Customising the messaging increases relevance, which will improve the response rate. As well as rewarding the loyalty of customers who are already highly engaged, marketers can also identify those on the borderline and present exclusive offers. For customers who buy in large quantity but irregularly, marketers can nurture the relationship with personalised content. Up-sell and add-on campaigns can reignite interest among dormant customers; cross-sell campaigns can increase the order scope among low-value customers.

Customer Retention Strategies

RFM analysis can highlight at-risk accounts or those showing signs of decreased engagement (e.g., low recency or frequency scores). Armed with this insight, marketing and sales can collaborate to implement proactive retention strategies: targeted outreach, account reviews, or special incentives, to prevent churn and maintain customer loyalty.

Cross-Selling and Upselling Opportunities

By adding insights about purchasing patterns by product category into the RFM analysis marketing and sales teams can identify cross-selling and upselling opportunities. For instance, accounts with a high frequency of purchases but lower monetary value may be targeted with upsell offers to encourage larger transactions or additional product/service adoption.

Preparing for RFM

Before we get into the hands-on, there's a context issue you need to be aware of. The whole point of RFM is to be able to identify segments in order to take action. That usually means an outbound communication or campaign, which has a relevant message and a Call-to-Action.

The analysis you're about to do is based on customer financial data, which is most likely at the level of customer account or ship-to address. When you execute the outbound promotion, you'll be wanting to communicate with individual people. The implication is that you'll need a rock-solid method for turning a list of companies into a list of contact names. That probably means extracting Account IDs as well as company names from the financial system.

Another point to be aware of, is that data volumes expand very quickly: 3.000 customer locations x 6 product categories x quarterly values x 5 years = 360.000 data rows. If you extract data at a fine level of detail you can always summarise at a less granular level. But if you extract data at a coarse level of detail, then drill-down is impossible. If you really do need more detail, you'll have to start with a new extract and analyse the data all over again. It's a balancing act. Think it through carefully and choose accordingly.

Extracting Data for RFM

That said, here’s a simple way to do RFM analysis in practice.

• First - extract data from the financial system. For the chosen period, you’ll need to extract a file showing each transaction as a row: the identity of the customer organisation; plus the date of the transaction; and the monetary value. For some customers, there will be multiple transactions.
• Next - condense this transactional information into a summary by customer. (In practice: start with an excel spreadsheet and make a pivot table.) This list shows every unique customer (name and ID), plus three additional columns of information: data of most recent purchase (use the function ‘maximum value’); the total number of purchases during the period; the sum of all purchases (count) during the period. Copy and save as a new table.

Now use this new table to create the composite ranking. Work through the characteristics one by one. Rank the customers and then create the labels. (Tip: for each dimension, give the most important sector the highest value of 5 and the least important the value 1. This approach simplifies visualisation: the most important segments will appear at top right in charts.)

• Rank all the customers in the spreadsheet by ‘date of most recent order’, from newest to oldest. Create a new data column to hold the labels called ?recency‘. Now, divide the ranking into quintiles and label each customer with the relevant value in the new column.
• Re-rank the customers by the quantity of orders, from most to least often and create a new column called ?frequency‘.
• Finally rank all the customers from largest to smallest and create a new column for the labels called ‘monetary’.

The next step is to decide the prioritisation of the three characteristics and then sort the entire table using the three label columns ‘recency’, ‘frequency’ and ‘monetary’.

In the final step, the combined ranking of three values for each of three characteristics gives (5 x 5 x 5) = 125 segments. The classic prioritisation is Recency then Frequency then Monetary. Rank the segment scores in descending order are: 5.5.5; 5,5,4; 5,5,3; 5,5,2; 5,5,1; 5,4,5; 5,4,4; 5,4,3; etc. In theory, each segment should contain 100% / 125= 0.8% of the total customer base. In practice, clumping occurs. About 5% of the segments will contain multiples of the theoretical average, while about 10% are completely empty.

What RFM analysis shows

Let's begin by simplifying three dimensions to just two: we'll ignore Recency for the moment and show Monetary on the vertical axis, versus Frequency on the horizontal. A quintiles chart of M x F comprises 25 segments. By using quintiles, we've set up the analysis so that each row or column holds 20% of the total number of organisations (71 of 355).

In theory, each segment represents 4% of the total number of organisations, but in practice, clustering occurs: the number of organisations in a cell varies from zero to a high of 53 (16% of the total).

And - precisely because "customers are not created equal" - there's an even bigger variation in the monetary value of the organisations within each segment. In the top-earning quintile (M5), 20% of the customers generate 68% of revenues; in the bottom-earning quintile (M1), 20% of the customers generate just 1% of revenues.

Examine the same data plotted as Monetary versus Frequency and there's also a very clear division in the customer base. The 40% of organisations in frequency quintiles 4 and 5 generate 78% of the revenues. The remaining 60% of the customers generate only 24% of the revenues. This is the classic Power Curve with the Long Tail in action.

A similar chart of Customers showing Monetary versus Recency shows a more even distribution of numbers of customers across the segments. Nevertheless, the distribution of Revenue in the Monetary versus Recency table confirms the power curve among both irregular and infrequent customers, as well as regular buyers: the M4 and M5 quintiles dominate the revenue table.

Here's the best bit: if you've analysed your data extract as a Pivot table in Excel, a swift double click on a cell will call up a list of all the customers in that segment. Now you can go search your online marketing database for matching contact names.

Other ways to define 'Return'

If we limit our view of success to mean revenue only, then there's a temptation to target every promotion to the same segments. Over and over. Repeat that too often and our marketing efforts will quickly become counter-productive. But the 'return' on a promotion doesn't have to be measured in revenue and profit.

What else can you do with RFM?

RFM analysis is a great way to create clarity that supports decision making. Here are a few examples:

Resource Allocation

The ability to identify segments accurately by value means that business managers can allocate sales channels, communications methods, and staff resources effectively. Business managers can set boundaries between in-house vs partner channels for example, to identify which high-value customers will be served by the direct sales force, and which by channel partners, distributors or agents.

They can also identify the lower-value segments that require low-cost communications and sales methods. For such groups, the combination of online marketing, outbound email and eshop can provide fast response with cost-effective distribution.

Similarly, the ability to accurately understand how segments fit into the bigger picture of customer service enables Sales teams can allocate resources more effectively, ensuring customers receive the attention they deserve.

Service Development

RFM analysis provides valuable feedback on the performance and appeal of services within different customer segments. By analysing purchasing trends and feedback across different customer segments, Service managers can identify areas for enhancement or innovation to better meet customer needs and preferences.

Segment-Specific Pricing Strategies

The insights from RFM analysis can also be used to develop segment-specific pricing strategies. For instance, premium pricing may encourage a customer to strengthen their commitment to their supplier; or to thank high-value customers who have already consistently demonstrated a strong loyalty. More competitive pricing or promotional offers may be used to attract or retain customers with lower RFM scores.

What you can't do with RFM

RFM is a highly effective way to understand history and to prioritise repeat selling to existing customers. On the basis of historic patterns you can find more people who 'look like' known buyers. It's great for cross-sell and up-sell.

If you're launching a new product within a known category, RFM can suggest candidates. But it will not necessarily help you launch a new category of product. Launch an existing product into a new segment is another area where RFM can suggest experiments. In both cases, though, careful measurement and analysis will be essential.

And when not to use RFM analysis? Where RFM is least likely to help is completely new territory without any history: new customer acquisition; new product categories; new industry sectors; new country markets. But don't let that stop you. Win customers, make sales, get experience, gather data. Then run an RFM analysis to trigger more ideas.