Application of RFM analysis in customer base segmentation

Segmentation and targeting are alpha and omega marketing, and email marketing is no exception. Even if you have a client base accumulated over the years of work, fan mailings can do you no less harm than good. You need to be able to send exactly those letters that your customers will need and will bring you the necessary sales or, for example, register on the site without filling up your customers' boxes with unnecessary messages that they would perceive as spam. To do this, you must clearly understand to whom and why you are sending each of your letters.



Especially for Unisender habrablog, we have prepared the “ Targeting and segmentation of email newsletters ” cycle. In the second article, we turn to the consideration of specific segmentation techniques and describe the use of RFM analysis in the segmentation of the client base.
')
In classical marketing, there is a technique of RFM analysis, which is used to predict customer behavior based on his past actions. RFM - an abbreviation of the words Recency (novelty), Frequency (frequency), Monetary (attachments).
Recency is the time elapsed since the client’s last activity, be it a purchase, a link, or just a letter opening.
Frequency - the number of actions performed by the client.
Monetary - client's cash costs for goods and services of the company.

It is assumed that a client who has recently manifested himself, shows increased activity from the moment of his registration or spends more money on your products will be more interested in your advertising campaign.

There is also a variation of RFD that is quite relevant for legal mailings, in which D (Duration) is the total duration of acquaintance with a client. For the method I have described below, it is not fundamental whether the customer’s investments or how long he has been a subscriber are taken into account.

During the division into segments of our address list, we consider each of the details of the puzzle separately and give the picture integrity at the end.

R = Recency

In accordance with the specifics of the services you offer, it is necessary to determine the criterion of customer activity. It will be the date of the last purchase, the last time you clicked on the link in the letter, or the last opening of the letter is enough - you decide.

It is necessary to conditionally define several time cycles for which activity will be taken into account. As a rule, when it comes to mailings, marketers stop at five periods of the last action: the last month, two months, three, six months and a year.

Now, using the collected data, you should “scatter” clients into temporary groups. You should have something like this:

1. 0-30 days: 5,100 buyers / subscribers
2. 31-60 days: 12,300
3. 61-90 days: 32,800
4. 91-180 days: 75,000
5. 181-365 days: 123,400

Group 1 hereinafter will include the most profitable customers. The “best” segment will always have the smallest number of buyers / subscribers, while the “worst” segment will be the largest. Do not try to change the criteria of groups so that they are equal. On the contrary, you should have something like a pyramid.

F = Frequency

Here we will consider how often the client has been active. As in the case of Recency, we will focus on five points, each of which will contain a certain quantitative range of purchases / actions.

Again, group 1 will take into account the most active. Think carefully about the number of actions your client is considered to be the most profitable: setting the bar too high or too low can knock down the counting accuracy.

The final version might look something like this:

1. 16+ purchases / actions
2. 11-15
3. 5-11
4. 2-4
5. 0-1

M = Monetary

This item defines the “quality” of the actions performed by the user, as a rule, measured in the money spent on your goods and services. It may also be indirect income from the actions of the client. In some cases, when the benefit from user actions is difficult to take into account in money or another specific value, or the value of each action performed by the user is the same, this item can be replaced by the client’s subscription duration or be absent altogether.

Just as in the previous paragraph, it is necessary to determine the individual limits for monetary investments that are individual for each company.

1. $ 1001 +
2. $ 601-1000
3. $ 351-600
4. $ 201-350
5. $ 0-200

Thus, we will get almost 125 groups: 555, 554, 553, 552, 551, 545, and so on up to 111. But this is far from being the case, but individual letters need to be addressed to each group: as a rule, this detailed separation is used for analysis. By grouping customers in this way, you immediately get a complete picture of what is happening with your customer base. For example, 5R-5F-1M customers could only contact you at one time, spending a considerable amount, but not counting on long-term cooperation. But 5R-3F-1M are more likely to become unhappy with the services of the company or have lost interest in it.

1R-1F-1M

This is the cream of your client list. If you correctly chose the limits of the groups R, F and M, then this segment should be extremely small - no more than 5% of the address base. Whatever you do, you can hardly ruin the relationship with such clients.

In this regard, many companies believe that it is better to leave relationships with such customers as they are - and they will continue to lead in the sales lists. But - this they lose the opportunity to give them lead by a large margin. There is no limit to perfection.

It is possible to expand the boundaries of cooperation with these people by establishing a loyalty program, inviting them to special events, or by questioning them regarding the wishes for the development of the company. It is important to show such clients in every possible way that they are respected and welcome guests, and not random passers-by.

5R-5F-5M

Although these customers seem to be the least promising, you shouldn’t completely discount them: at least once, but they still showed interest in your products. Most often, advertisers are preparing for them special, “provocative” messages that allow them to get rid of the ballast from those who do not show interest at all (or simply abandoned an electronic mailbox) and transfer the rest to the next category.

“1” in only one category

Those who have only one unit in the category of Recency, should be given a little time to decide. They know about your services and may soon show interest in them. Appetizing mailings in this case is the most effective way to keep their attention.

Customers who buy frequently (1F), but for small amounts, are valuable to you by their constancy. Try to use the technique from the previous article: offer them products that are usually bought for the company with those that they purchased. Perhaps this can expand the range of their purchases.

A client with 1M is extremely priced for your company with its substantial investments, but is not active. It should show its special value to you. Take a look at what they bought and how long ago they didn’t take any action. You need to carefully figure out what they would like from your company? What they lack to become your regular customers? A small survey can be beneficial.

"5" only in one of the categories

These users are a promising segment and space for your research activities. They are constant enough for you to experiment and find a suitable way for your company to pull customers from 2-4 in each of the points.

The RFM methodology is far from absolute, but extremely useful as an address list analysis tool. Look around: having done relatively little work, you already see an individual approach to your clients. Things are easy: get down to action.

Based on Kelly Lorenz articles - RFM Segmentation and Analysis: Part 1 , Part 2 , Part 3 .

" Targeting and segmentation of email-newsletters ", especially for the habrablog of Unisender . Previous article .

In order not to miss the following materials on this topic, click the heart in our profile and mark the blog of the company Unisender in the settings of your habralenty .