Customer segmentation is an obligatory measure in internet-marketing. Only knowing what customers type form the database and what kind of offer they can be most likely interested in, you can count on success of your advertising campaign. The issue of customer base segmentaion is especially topical in the context of e-commerce development. The number of internet stores grows but only a few of them benefit from online commerce. Your customers segmentation by certain parameters can help understand their preferences. One way of customer list segmentation is RFM-analysis.
Features of RFM-analysis
RFM-analysis is a possibility to evaluate the customer base for its propensity to response to the offer. Roughly speaking, such segmentation will help understand which customers will make a purchase and which ones will not even open the letter.
In this way you can determine how appropriate is spending advertising budget on this or that customer attraction: from the purchase history you can decide if the customer is worth new marketing strategy. It is called target selection, target strategy.
Get lost in the search results?
Let Promodo experts help your website index better by optimising your pages, improving your content, and enhancing your linking strategy.
By clicking this button you agree to our terms & conditions
The analysis is based on three variables:
- Recency ― recency of the purchase (time period from the moment of the last purchase until today)
- Frequency ― frequency of the purchases (the total number of customer’s purchases for the whole history of dealing with the company)
- Monetary ― the monetary value of the purchases (total amount of money spent on the purchases)
The sequence of letters which comprise RFM abbreviation is not random, it shows the degree of importance of each variable. Recency is more priority criterion comparing to frequency and value; consequently, value is less priority index than frequency.
You should not consider RFM-analysis as customers’ behavior prediction tool. With its help you can more or less accurately forecast this behavior. Analysis starts from purchase history. In other words, RFM-analysis can be applied only with the established customers, as potential customers do not have a history of purchases yet. This method of segmentation of customer base is equally suitable both for consumer market and for the corporate one.
The analysis is implemented on the base of exact quintiles method, which comparing to hard-coding method is more accurate.
Segmentation by recency of purchases
Where shall we start from? There is a customer base where the demographic data on each customer and the history of the purchases are stored (What? How many? When? At what price?). First the database is analyzed by recency of purchases. To do this all customer base must be divided into 5 accurate quintiles depending on when the last purchase was made.
We get a sort of column divided into segments. The top one is given code 5, the next one – 4, then – 3, 2 and 1.
The segment with code 5 includes buyers who made a purchase not long ago, that is, more recently than the others; segment with code 1 includes those who made a purchase long ago, that is, earlier than the others. The rest of segments (2, 3, 4) are the data on the customers who made their purchases during the period between the earliest and the most recent purchases.
To make the result of the segmentation more accurate you should decide what time period should be taken as a basis. In other words, “the most recent” means how long ago?
Segment “5”: 0-1 months
Segment “4”: 2-4 months
Segment “3”: 5-8 months
Segment “2”: 9-11 months
Segment “1”: 12 months and more
We’ve made database coding by recency. Based on the results obtained we can assume what segment representatives will respond to our offer more readily and faster. They will be the customers with code 5. The level of response in quintile 5 will be higher than in quintile 4, which, in its turn, will be more active than representatives of quintile 3 and so forth.
Why is it so? The fresher the memory of the purchase made recently, the more probability of making one more purchase. In this case psychology comes first. What does a purchase means for the person? Satisfaction from right investment of his money, from acquisition of the long-awaited thing which will make his life better. When pleasant emotions from the purchase, made, for example, yesterday, are still fresh, there is a big chance to use them and sell something else to such buyer. And vice versa: the longer time passed since the last purchase, the less probability that this customer will make the next purchase in the same store.
Segmentation by frequency of purchases
It should be noted that for each business area the term “purchase” has different meanings. It can be purchase of a TV-set (if we speak about internet store of audio, video equipment), or transition to a new tariff plan (if we speak about internet provider, web-studio etc.) and also any other change of previous model of cooperation by the customer. We should decide on understanding of the term “purchase”.
Like in the previous scheme we should segment customers/subscribers into 5 quintiles. Quintile with code 5 will include customers who make purchase most often, quintile 4 will include customers who make purchases not so often as the representatives of the higher quintile, and so on in the descending. You should understand that those who make purchases most seldom and have a short history of cooperation with the company will occur among customers of quintile 1.
For example, a customer made a purchase yesterday. By recency criterion he will get code 5, by frequency criterion – 1. Such customers who made one purchase both yesterday and a year ago are many. That is why quintile with code 1 will be big. Segmentation of customers by frequency of purchase will dramatically demonstrate customers’ “affection” to the company. The more often the customer uses services/offers, the more probable high level of his trust.
In order to get the most accurate result of coding of database by frequency you should decide on what frequency will be optimal, what should be the landmark, what number of deals (sales/orders/connections) will be that maximum to start from.
Segment “5”: 20 and more purchases
Segment “4”: 15-19 purchases
Segment “3”: 11-14 purchases
Segment ”2”: 5-10 purchases
Segment “1”: 0-4 purchases
Segmentation by monetary value
The criterion of monetary value of purchases is less important than the criterion of recency and frequency. However, it should also be considered. In order to divide database by monetary value, you should determine for each customer the sum of money he spent on purchases (other actions) on the site. Customers who spent more money than others will enter the segment with code 5, those who spent less than others – segment with code 1.
But you should understand that customers of the lower segment (1) can buy more often but for less money. That is, they spend large sums of money not willingly; they will better buy more not very expensive products than make more serious purchases. Thus, you can assume what segment should be offered what products (at what price and in what amount).
Segmentation has been made. What next?
The results of the analysis must help make out what categories of customers you are about to deal with and how to use this information:
- Who spends much money and often?
- Who makes expensive purchases once in half a year?
- Who visited the site long ago and who can be “reanimated”, and so on.
By segmentation each customer is assigned a three digit number consisting of recency, frequency and value indexes. Total number of cells is 125, they look like 555, 554, 553, 552, 551 … 113, 112, 111. These are customers’ features expressed in numbers.
In order to understand how well the analysis is made, you should formulate an offer and send it to the test-group – that is, make a list of recipients according to the principle “every N-th”. As a rule, every N-th – is every 10-th (other selection is also possible). The results of response to the offer can be used when determining the level and index of break-even.
The task is to make profit, not to incur losses. In order to control this process the notion “break-even” is introduced in marketing. It is a percentage ratio of profit from selling to the test group to the cost of advertisement, addressed to the same group.
BreakEven = cost of the act of sale / net profit from one sale
For example, if you could earn 100 dollars from one sale and cost of sale comprises the same 100 dollars, it means that the break-even is 0. That is, 0 is the index which the test group should reach to provide the break-even of the advertising campaign.
In order to understand which of the cells is the most profitable and vice versa we need a break-even index. It is also calculated according to the formula and demonstrates how break-even is this or that cell. Thus, if the index obtained will be with minus, it will show unprofitability of RFM-cell, if it equals 0 – the cell does not yield loss.
The formula to determine a break-even index:
k = ((r – BrEv) / (BrEv)) * 100%, where r ― is the level of a cell response.
The size of the cell is the criterion which is important for reliability of the results. Depending on whether we are speaking of consumer customer markets or corporate ones, the size of each RFM cell will vary. The more the number of customers, the bigger is the cell, and vice versa. For corporate customers, let’s assume, these sizes will be not big. On the one hand, each cell must have rather many respondents in order that data was as reliable as they can be, on the other hand – rather few, so that the cost of marketing activity was not very high.
The following formula can help determine the minimum volume of the cell:
RFM-cell = 4 / BrEv
What will we benefit from knowing which RFM cell has which code?
- have recently made their latest purchase
- often make purchases
- spend much money on purchases
This niche is the most attractive. The main emphasis in this case should be made on forming loyal relations with such customers. And vice versa – offers of discounts, actions and sales should be given up.
- the latest purchase was made long ago
- purchases are made very seldom
- little money are spent on purchases
Such customers are not of great value. In some areas of business it would be more appropriate to abandon them, than to spend budget and experts’ time to draw their attention. It can well be a “dead” zone, a segment of “transit” customers who don’t care where to buy and who not very easily part with their money.
- have made a purchase recently
- don’t buy very often
- don’t spend much money
These can be:
а) customers who made a purchase in the store for the first time
б) customers who restored their interest to the internet store. At first glance such customers are hopeless. On the other hand, there is a chance to interest them by attractive offer sent at a proper time. It is important to motivate them by discounts, action and bonus offers.
- made the latest purchase long ago
- seldom make purchases
- spend much money on purchases
Such customers, most likely, make their purchases very deliberately and carefully and are ready to big investments at right motivation. The fact that the purchase was made long ago testifies only to the fact that your actions on bringing these customers back must be very energetic.
RFM-analysis can serve the marketing manager right, if the goal is to study the customer base and understand what, whom, when, in what volumes and at what price should be offered. Eventually, it is one of the ways to earn more, using the old resources.
About the author: Tatyana Gavrilina, specialist in e-mail marketing at the Promodo Company. The author of a number of publications on internet marketing, in particular – SEO copywriting and promotion in social media.