Practical use of RFM customer segmentation

Evaldas Mdata, segmentationLeave a Comment

Do you perchance mean that these things have a substantial monetary value to you?’ ‘That’s right.’ ‘Oh,’ Marlowe said beaming ‘in that case, Mr. Carter I am fully prepared to meet you on your own ground.’ ‘Come, come!’ Do you keep a diary?   I’m going to ask you a question straight from the shoulder!’ ‘Bring it!’ ‘What does RFM really stand for?’

RFM (Recency, Frequency, Monetary) is a data backed segmentation variable in marketing that measures the loyalty of customers. It groups customers based on their purchasing history – when, how often and how much did they buy.

RFM improves customer segmentation by dividing them into different groups to identify and target customers who are genuinely interested and likely to participate in promotions and future personalization services.

RFM model example

Mapping customers dependent on a single parameter is inadequate because it may undergo a change. Your best customer today can quickly become a marginal customer tomorrow. Then giving details of how much money the company has earned in the previous year, you have to explain to the boss tomorrow why you have lost so many customers.

You can say that people who are heavy spenders are your best customers. Maybe, but then again maybe not! What if they bought from you only once or infrequently? What if they were unable to continue using your product, then what? Can they still be considered your best customers? I don’t think so.

That’s why, RFM model depends on three customer attributes ordered by importance. This allows the organisation both to focus on the segments that best fit the customer profile and to avoid customers with less buying potential.

RFM is a handy technique to identify your high value customers, understand behavioral data and deploy more triggered and personalized email marketing campaigns to increase sales, maximizing long-term customer retention and lifetime value.

If they bought once, not so long ago, this shot gives them higher points. If they bought many times, they hit higher score. And if they spent bigger, they win the match by getting more points then. Save and combine these three attributes to create the RFM scoring opportunity and encourage customer to buy more per purchase occasion.

Now you can penetrate right into the core of segmentation and partition your customers into clusters using these Recency, Frequency, and Monetary scores. Here is an example for you of different types of customer segments that you can create with our RFM analysis. Note that each segment is denoted by scores which let you find those customers. We’ll go into more detail how are they calculated in the next section.

Segmented customer personas

  • Best customers – 333 – Bought recently, buy often and spend a hell of a lot of money. Make them feel important and appreciated. Market to them with caution, without price incentives, do not squeeze on a high profit margin. Tell them about launch of new products, incorporating the newest ideas and the most up-to-date features, how to connect on social networks, and about branded loyalty programs, because they are promoting your brand.
  • Loyal customers – X3X – Your most regular, loyal and reliable purchasers. Have a high rate of growth in Frequency but may or may not remain constant in the Recency and Monetary Value scores. Always keep coming back for more, easily sign up for loyalty programs that appeal to them, responsive to promotions, just politely remind them. Ask relevant questions when they leave reviews to help yourself with SEO, strengthen the bonds, and improve the health of the brand.
  • High spenders – XX3 – Very wealthy, maybe a residence of the dukes or something, they have a lot of money, and unable to stop themselves spending it over the lifetime of your relationship. Usually content themselves with a few big purchases, or with a few small ones. Motivated enough with your products or services to spend a lot more later. Will probably participate and hold membership in several loyalty programs. Target them with the most expensive high-end items.
  • New high spenders – 313 – Have made significant purchases on their first buying experience. To increase the amount, and to convert them into returning customers, provide an incentive offer for a second purchase using triggered email, ask about customer experience on your website. Provide the right and effective onboarding support, build relationship.
  • Low loyal customers – X31 – Buy often but are not able to spend more than they can afford or more than they think something should cost. Make purchases carefully, but trust your brand, all you have to do is to achieve an increase in their spending for each regular purchase. Would be perfect to offer both rewards and receiving discounts off of a certain amount of spending (eg. “Spending $100 gets you a $15 credit).
  • Rare high spenders – X13 – Don’t buy so often, but when they do are likely to spend quite a lot. Lay out a large amount of money but decide to buy occasionally. Set up a much higher price target for them getting them to acquire more profitable items. May choose to gives you a seasonal break, so leverage their seasonal buying behaviors and enroll them in your seasonal membership. Incorporate seasonal marketing strategies into marketing plan and remind of your brand when they re-engage.
  • At risk customers – 233 – Valuable customers that purchased frequently and spent big money, but some time ago, at risk of being lost for good. Create and implement moderate and personalized retention strategy with more competitive pricing, new product launches, send emails to reconnect, reach out.
  • Churned customers – 133 – Best customers that were lost. Requires aggressive price incentives and reactivation campaigns. You can also send a customer survey and get the feedback on why they left: poor experience, seasonal products or a ‘one-time’ thing. Revive their interest with promotional incentives of more aggressive discounts off a range of products based on past purchases. Don’t loose these customers to cheaper rivals. Recreate brand value. Win them back!
  • Churned cheap customers – 111 – Spent as little as possible, bought very few goods, all orders from a long time ago. Extremely unlikely source of repeat purchase, doesn’t really worth time and trouble. Best bet is to re-confirm these customers incorporating them into your email newsletters, maybe an email will bring them back.

RFM Analysis Example

In order to calculate RFM metric, and apply it to your database you’ll need to have each customers orders. Then using the details we get: (R) days since last buy, (F) total of transactions, (M) average order value

Customer ID          Name Recency (days) Frequency (times) Monetary (aov)    Segment
1 Dwight Bleichert 30 4 135 Low loyal customers
2 Philip Marlowe 40 5 185 Loyal customers
3 Jack Carter 300 1 30 Churned cheap customers
4 Bob Lee Swagger 100 1 340 Rare high spenders
5 Sam Spade 60 4 600 High spenders
6 Buzz Meeks 210 5 140 At risk customers
7 Lee Blanchard 17 1 420 New high spenders
8 Bud White 10 9 150 Best customers
9 Jessy Saunder 280 4 220 Churned customers

Consider customer Dwight Bleichert – he last ordered 30 days ago and had a total of 2 orders worth $135 until now.

Implementing RFM Score Analysis

Once the sequence of events is reconstructed and RFM point values from the actual purchase behavior are assigned, each customer is a assigned a score of 1 to 3 is for the recency, frequency and monetary values of the purchase data. 3 is the highest possible value, and one is the lowest. RFM values and RFM scores are not the same thing. Value is the actual value of R/F/M obtained for each customer, while Score is a certain number from 1-3 based on that value. A final RFM score is calculated simply by correlating individual RFM score numbers.

To determine the total score, we need to sort an arrangement of values in ascending and descending order of pitch to find out which values come top or bottom. So we have 8 customers and only 3 scores, so we assign a score of 3 and 2 to the first six records, the other two get 1 repeating each digit a number of times, in order to facilitate the subsequent filtering of unwanted noise. Since RFM score depends on recency, frequency and monetary value ratio, the real power of the analysis comes from concatenating of the three digits R, F and M.

Lieutenant Marlowe closed his eyes, first mentally viewing the different scenes, then correlating the data…

CID R Value R Score CID     F Value F Score
8 10 3 8 9 3
7 17 3 2 5 3
1 30 3 6 5 3
2 40 2 1 4 2
5 60 2 5 4 2
4 100 2 9 4 2
6 210 1 7 1 1
9 280 1 4 1 1
3 300 1 3 1 1
CID M Value M Score          CID   RFM Score
5 600 3 1 321
7 420 3 2 212
4 340 3 3 111
9 220 2 4 213
2 185 2 5 223
8 150 2 6 131
6 140 1 7 313
1 135 1 8 332
3 30 1 9 133

RFM model calculation on scale of 1-3

So, customers who bought recently are your repeat purchasers, and businesses thrive on them, because they spend a lot, that’s why they are assigned score of 333 – Recency(R) – 3, Frequency(F) – 3, Monetary(M) – 3. Now on the other side of them from you the customers that spend the lowest, hardly surprising with any purchase at all, and that quite some time ago – a score of 111. Recency(R) – 1, Frequency(F) – 1, Monetary(M). Customers in-between with frequent purchases they also spend significant amounts of money in your shop.

Now, we are no contemporary avant-garde composers, obviously, so why concur in the principle of the triad for each score?

Methods of RFM calculation for ranking the RFM values on the scale of 1 to 3 will depend on various different factors like the type of business, the stage that business is at, and the nature of the value proposition, but as an example we’ll show the most common methods:

Quintiles, five equal groups into which a population can be divided according to the distribution of values of a random variable.

The term is from maths as well as these ones:

Percentile, each of the 100 equal groups into which a population can be divided according to the distribution of values of a particular variable.

Tertile, either of the two points that divide an ordered distribution into three parts, each containing a third of the population.The first tertile results include January through April’s revenues.

They are basically the same and interchangeable, but instead of dividing the data in 100 parts, we divide it in 1-3 parts. By using 3 parts we can get up to 3x3x3=27 unique segments total. To divide them further we can use following categories:

  • Recency – recent, lapsing, lapsed
  • Frequency – one-time, repeat, loyal
  • Monetary value – low, medium, high

Note there is one caveat for frequency when working with real life data. You usually want to pick it yourself instead of just slicing into 3 parts. This is due to the different behaviors between one-time vs repeat vs loyal customers. Usually there is disproportionally (70% on average) more one-time buyers than repeated ones which would make 1 time buyer be included in 2 parts. Thought this kind of behavior could be different for different ecommerce business therefore it’s always better to pick the values before hand.

So here’s what it might look like to enter a somewhat complicated mathematics, but then again it exhibits the structure associated with this type of proof used to solve many problems that involve proportions, without mentioning the word “proportion”.

Tertiles work for a broad range of businesses, integrate with a wide range of systems, giving optimal flexibility and exceptional share data for any of a range of purposes. The values of the variable being ranked e.g. customers are evenly spread without skew data.

Tertiles is a highly recommended method of calculating RFM score, it is almost universal across all products and services, all industries and all types of customers.

We make use of tertiles to provide our own RFM segmentations at StackTome – key insights into merchant business performance through online tools, and data-driven marketing analytics solutions for online retailers. Just take your critical information assets such as customer data, and give it a score from 1-3 to R, F and M values.

Business Application of RFM Customer Segmentation

RFM segmentation is an ‘old’ concept going back to catalog retailers in the 90s who used mail for receiving and sending orders. The idea was to be more efficient when it comes to marketing to specific segments of their customers as sending ads/offers by mail is not cheap. And while we are still waking up to this power, the biggest selling brands have used RFM based segmentation to maximize an impact upon the return of investment of retention action and marketing campaigns for years sending targeted messages to those segments.

RFM and CLV

New high spenders, Rare high spenders, At risk customers, Churned customers are 4 segments that you need to pay particular attention to. By sending personalized emails, getting the right content and highly relevant and personalized product recommendations make reconnections with these customers. You may even capture their interest with exclusive access to sales and discounts before they are publicly available. Surveys offer valuable insights on what customers like or don’t like about your products and services and what make them buy from you, just run them! Identify upsell-cross-sell opportunities and brand’s offerings for segments that are more likely to respond, reduce customer churn. Optimize the marketing message, increase brand loyalty and word of mouth referrals, encourage them to add more items to their cart, and purchase high ticket items where appropriate, but don’t overdo it! Make them feel like they are receiving a meaningful advantage.

The second is the rigidity of attitudes indicating the presence in the game, which is the quality of staying firm in your commitment to brand consistently rebuying and repatronizing a preferred product despite extrinsic stimuli like marketing efforts. At times, a customer can be behaviourally loyal without displaying loyalty attitude (High spenders, New high spenders). And vice versa, a customer can display loyalty attitude without having loyalty behavior (Rare high spenders, Low loyal customers, At risk customers). Or they can be just loyal repeat customers (Best customers, Loyal customers).

RFM minimizes marketing costs and improves ROI

Direct impact on improving your marketing ROI and customer lifetime value, higher engagement rates, loyalty and conversion, greater success in new product launches and remarketing campaign, higher profitability, it’s all about RFM.

It’s costly and ineffective to waste time on untargeted marketing campaigns, and can even be detrimental to sales. Conversely, focusing on a smaller segment based on intelligent analytics will significantly reduce marketing costs, while also increasing results.

The roots of RFM can be traced to direct marketing. Where they eliminated unnecessary overhead costs of printing and shipping catalogs by targeting only those customers that were more likely to respond positively to the marketing campaigns.

So whether you are doing digital marketing, print or media, this type of segmentation is beneficial to reduce your costs, improve return on investment and remain competitive.

RFM for remarketing and retargeting campaigns

Remarketing is a technology that enables your previously browsed product to follow people across multiple websites in banner ads, in an effort to re-engage those who’ve have left your website without buying. This improves click rates, effectiveness and completing a goal conversion.

An easy way to use remarketing with RFM can be to export specific customer segments that can prove beneficial for your business in future – especially the New Customers or Promising ones – to campaign management solution like Facebook Audiences that streamlines sales workflow and promotes quick and organized tracking and updating to aid sales teams. Then show promotions to that group of people and continue slicing and dicing of members of a remarketing list, but that’s another story…

Summary

Use RFM model for high value customer identification or develop more targeted personalization campaigns using RFM analysis. Use Stacktome as an in-depth marketing automation solution, if you are an online retailer. Run comprehensive reporting, automated email and outreach campaigns based on RFM. Stacktome provides comprehensive RFM analysis, and many other business analytics and reporting tools.

Here’s how it looks https://www.youtube.com/channel/UCqdPTBcDZREUMlK1cpKDBjw/videos

‘Don’t be so impatient,’ Carter said. ‘And suppose you and I both try this for size?’ He deftly snapped one handcuff on Marlowe ‘s wrist and the other on his own. Detectives have to discover the opportunity R, the motive F, the means M and keep these three balls in the air.

Leave a Reply

Your email address will not be published. Required fields are marked *