Customer segmentation, main principles and use cases

Summary:

  1. Definition and principles of customer segmentation
  2. Segmenting your customers: where to start?
  3. What means should you use to segment your customer portfolio?
  4. Top list of the best customer segmentations
  5. Interest of the segmentation for the customers and the company
  6. Targeting or segmentation?
  7. How to know which segments to target?
  8. What can be personalized with customer segmentation?
  9. Prédictive segmentation
  10. Dynamic predictive segmentation
1. What is segmentation? Definition and main principles

Segmentation consists of dividing customers into homogeneous groups based on their profiles or behaviors: these groups are commonly called “segments”.

Customer segmentation has two purposes:

  • the first objective is customer knowledge: by segmenting your customers, you can determine the proportion of students, working people, retired people, urban people, rural people, couples, etc. You can identify how many customers are loyal to your products, who are your potential customers, who are at risk of leaving your company, who prefer to be contacted by text message rather than by email, etc., and you can measure the share that each of these segments represents in your sales figures
  • the second purpose is the personalization of the customer relationship: indeed, personalization creates commitment, loyalty and customer value. To personalize marketing actions, each segment is addressed differently. The customers of the same segment receive content that is relevant to them and that makes them react in the same way.

This is why customer segmentation is the keystone of any marketing strategy, since it not only provides a detailed view of the customer base and its composition, but is also the indispensable tool for personalizing customer relations over time, according to the expectations of each targeted segment

“Personalizing the relationship drives customer engagement.
Customer engagement breeds loyalty.
Loyalty increases customer value.”

Thus, with a well executed segmentation, you not only improve your customer knowledge, but more importantly you create engagement and value.

Before embarking on segmentation, make sure you have a single customer vision, or a solid UCR (Unique Customer Repository). If your information system automatically considers that 3 purchases = 3 customers, it is because it is not capable of recognizing a single consumer who buys several times: prioritize the UCR project before any segmentation project. 


2. Customer segmentation, where to start?
Une image contenant personne, homme, table, portable

Description générée automatiquement

In order to segment your customers, you must first determine the right segmentation criteria to use. Before going any further, let’s see what a segmentation criterion is. 

To segment your customers, you will divide them into homogeneous groups, i.e. “segments”. All the customers in a segment share some common points:

  • Among the simplest are the commonalities of profiles, such as region of residence, gender, age;
  • more complex commonalities, based on customer behavior: for example, customers who made their last purchase less than three months ago, or customers at risk of leaving.

These common points are called segmentation criteria: they are the keys to dividing your customers into each segment.

The segments you will create can be mono-criteria: for example “my customers under 30”. Or they can be multi-criteria: “my customers under 30 years old registered to my loyalty program”.

In an operational logic, you will compose your segments in such a way as to isolate your groups of customers who react in the same way: you will thus address each of your segments with specific offers and messages. For example, you will address your customers differently depending on whether they belong to the “loyal” segment (a segment that permanently needs privileges and recognition) or whether they belong to the “at risk of leaving” segment (a segment that requires a strong punctual action).

For this, there is no need to compose fifty segments: for daily use, five segments can be enough, as long as they are well constructed, profitable and relevant to your business.


3. What means to use to segment customers?
Une image contenant personne, portable, intérieur, table

Description générée automatiquement

Beginner level:

If you are new to segmentation, and your customer information is recorded on Excel-type tools or even paper files, you can already start by listing your customers according to simple criteria: gender, age, social class, place of residence. To these different customer segments, you will offer personalized items.

For example, a high-income customer will be offered a superior women’s range rather than a first-price men’s item.

This approach can be done manually.

Intermediate level:

You have a usable customer database, but limited resources: using queries (SQL for example), you can already segment your customer base once or twice a year according to more advanced criteria, for example RFM. 

RFM segmentation (Recency, Frequency, Monetary) is as tedious to set up as it is robust: it is the essential step in marketing segmentation. With this segmentation, you identify the following different segments:

  • R segment: recent purchases, few purchases and small amounts > low potential customer
  • F segment: frequent purchases for small amounts, but no purchases for some time > small loyal customer at risk of leaving
  • M segment: few purchases and high amounts, no recent purchases > potential customer
  • R&F segment: frequent and recent purchases, for small amounts > small loyal customer
  • R&M segment: recent and high value purchases, but infrequent > potential customer
  • F&M segment: frequent and high value purchases, but no purchases for a while > very good customer at risk of leaving
  • R&F&M segment: frequent and recent purchases for large amounts > very good customer

These different segments, even if they are not used on a daily basis, have the merit of giving an initial vision of the business based on the purchases made, and of addressing customers according to their past purchasing behavior.

Advanced level:

You have one or more customer database(s) and intelligent segmentation solutions. Using these solutions, you address your customers according to predictive segmentations (we will come back to this later). These solutions are natively based on Artificial Intelligence: using algorithms and scoring techniques, they anticipate the future behavior of your customers based on their past behavior, or the past behavior of other customers who are similar to them. 

For example, the hotel industry uses Datacook’s predictive segments to identify at any given time their loyalty program members with the highest future value (i.e. with the highest revenue potential over the coming year): these customers benefit from a specific VIP program, which ensures their commitment to the brand and the revenue associated with this segment.

Omnichannel segmentation – Make sure your marketing segmentation reflects your omnichannel nature: unless you are a pure player, your customers are not only on the web! Your segmentation must integrate the behavior of your consumers over time on all your channels: on your website, in stores, in call centers, on social networks, on the mobile application, in agencies, etc.  


4. Top list of the best customer segmentations

To help you see clearly, here is the top list of the best customer segmentations, used by the most successful marketing departments. You can have your internal data scientists calculate these segments, or for a more operational use, use intelligent segmenting solutions, like Datacook.


  • SEGMENTATION BY CUSTOMER VALUE

Future customer value segmentation is based on predicting the future revenue generated by each of your customers. Calculated using AI, it isolates your best future customers, i.e. those with the most potential to generate revenue with your brand. 

The gambling industry, for example, relies on this segmentation criterion to optimize marketing investments according to the future value of its customers: in other words, marketing expenses are reduced on the low future value segment, and the investment is massive on the high future value segment, which ensures that all marketing actions remain profitable. – see the white paper on customer value.

  • SEGMENTATION BY LOYALTY

This type of segmentation is based on the loyalty rate of your customers: the loyalty criterion allows you to identify customers who are loyal to your brand and your products. 

Loyal customers need recognition and privilege, on a constant basis. In the tourist accommodation sector, for example, this segmentation is widely used to specifically address loyal customers: with upgrades, dedicated reception, etc.

  • SEGMENTATION BY CHURN 

On the other hand, customers at risk of churn are also the focus of attention: in the mutual insurance sector in particular, where competition is accelerating with the new law on intra-annual termination, this segmentation is strategic. 

Churn risk segmentation is therefore widely used to detect the weak signals of a potential customer’s departure, in order to retain them before it is too late. Since a customer costs significantly less to keep than to win, this segmentation is one of the most coveted in the industry.

  • RFM SEGMENTATION

The RFM segmentation is ultra-classic in marketing. It is the basis of all segmentation, as basic as it is robust: R for Recency of purchase, F for Frequency of purchase, M for Monetary, i.e. amount of purchase. Almost all companies still use this segmentation.

This segmentation provides a sort of three-dimensional view of your customers’ behavior, i.e. those who buy often for small amounts, those who buy once for a large amount, those who usually buy regularly but have not bought for a while, etc. This is a good first step to identify your loyal customers, and those who are likely to leave.

  • SMALL-MEDIUM-BIG SEGMENTATION

This segmentation segments customers according to whether they are Small, Medium or Large. This segmentation is certainly based on a past vision, but it offers an efficient visualization of the concentration of the business: what part of my turnover does my biggest customers generate? How much do my small customers represent in my annual turnover? 

This segmentation can be used by the Marketing Department as well as by the General Management.

  • SEGMENTATION BY PROFILE

This segmentation actually combines several different criteria: notably age, gender, social class, region of residence, household composition, lifestyle, equipment (car, energy), type of housing, etc. For B2B, it provides other information such as the status of the client company or its financial health. It is particularly useful for customer knowledge purposes. 

This segmentation is sometimes used by sales people on their prospects, in order to identify among them the best targeted ones in relation to their activity.

  • SEGMENTATION BY PRODUCT APPETENCE

Product appetence segmentation identifies the customers most likely to buy a particular product: it is used for cross-sell and up-sell, and allows to offer the right product to the right customer.

E-commerce uses segmentation by product appetence to identify, in real time, which product to push to which user.

  • SEGMENTATION BY CHANNEL APPETENCE

Segmentation by channel appetence allows you to communicate with your consumers on their preferred channels: sms, email, paper mail, phone calls, application, etc.

The mail order sector in particular, which is in the midst of digitalization, uses this segmentation to find out which of its customers have gone digital and which still prefer to receive a paper catalog. 

  • SEGMENTATION BY LIFE STAGE

Dividing its clientele according to the moments of life is a trend that is confirmed. It is used to anticipate the passage into retirement, the expansion of the household, the move, or other events at major stages of life.

Home-related companies (alarm systems, furniture, home insurance) use this segmentation to identify their customers at risk of moving: they thus anticipate the risk of contract cancellations and reduce their churn, or conversely they propose offers adapted to their customers’ future home.


5. What is the interest of segmentation, for the customers themselves?
Une image contenant intérieur, personne, femme, table

Description générée automatiquement

More than ever, customers expect a personal and lasting customer relationship with the brands they frequent, and the global containment episodes have only reinforced this trend. 

Customers who are highly loyal to a brand will appreciate being recognized as such, and benefit from privileges associated with their status, or even simply from messages explicitly acknowledging their loyalty (“dear Mr. X, you are one of our most loyal customers”); this recognition creates commitment and maintains the loyalty of this category of customers, fuelling the virtuous circle of the customer relationship. On the other hand, they will be extremely disappointed to be treated as ordinary customers, and this weakness can cost you their losses.

A brand’s ability to personalize its relationship is a powerful driver of customer engagement, as it meets a real demand from consumers.

“91% of customers are more likely to consume from brands that are able to recognize them and provide them with relevant offers.” – Accenture Personalization Pulse Check study.

And for the company, how is the benefit of segmentation measured?

The results of effective segmentation are measured in the short and medium term:

  • In the short term, it significantly improves the ROI of your marketing actions, especially if the segments are built in an efficient and intelligent way,
  • in the medium term, it extends the customer’s life and increases customer value.

6. When do we talk about segmentation, and when do we talk about targeting?
Une image contenant personne, extérieur, sport, tir à l’arc

Description générée automatiquement

Targeting is specific to a campaign: for each marketing campaign, you target one or more customer segments. 

A parallel could be drawn with archery. During the segmentation stage, you develop several targets (your segments): one target made only of straw, another made only of foam, and a last one made only of wood. These targets will be used for all your sessions (your marketing campaigns), for several months or even several years. When you send your arrows (your marketing actions), you aim each time at a particular target (a segment), so that the customers of the same “target” receive the same “arrow”.

Just as in archery, the segments are built once and for all, and are only challenged and modified on rare occasions: the targeting, on the other hand, changes according to the customers to be reached on each campaign and the marketing strategy.


7. How to know which segments to target?
Une image contenant intérieur, alimentation, grand, table

Description générée automatiquement

It all depends on your marketing strategy, and your budget.

If your marketing budget is limited and you have segmentation solutions, play it strategic: just animate your best future customers segment, those who will generate tomorrow’s business. Among them:

  • which ones are loyal? => reward them, recognize them 
  • which ones are at risk of departure? 
  • which ones are palatable to your product families? => push them the right products
  • which ones prefer email, phone, mail? => use their preferred channels

If your marketing budget allows you a wide range of actions with all your customers, then use five to ten strategic and relevant segments for your activity, which you will animate regularly. These segments must have a high ROI, as their profitability can be measured directly from your segmentation solution.

On the other hand, build sub-segments dedicated to specific uses: these will be more refined and will combine several key criteria.

In a general way, define your actions according to the position of your customers in the life cycle (“new”, “loyal”, “at risk”, “gone”, “revenue” segments), according to their value, according to their profiles, according to their product appetence, their channel appetence, and cross-reference these segments to increase the performance of your actions tenfold.


8. What can be personalized with customer segmentation?

Absolutely everything. It’s precisely the entire consumer experience, personalized from end to end, that will influence customer engagement.

Customize your marketing campaigns, your commercial offers, your web interface, your advertising content, your product recommendations, …and even your prices: B2B players segment their clientele (according to company size, for example) to adapt their pricing policy.


9. Sintelligent segmentation: predictive segmentation
.

Predictive segmentation consists in anticipating customer behavior. Artificial intelligence, which is essential for this segmentation, works in the following way: by matching data, both “hot” data (data relating to immediate behavior) and “cold” data (data relating to long-term behavior), it detects weak signals of future behavior, based on the customer’s past behavior or the past behavior of other similar customers. 

Since AI relies on learning processes, the larger the data history, the higher the reliability of the prediction.

Thus, the relevant combination of Machine Learning algorithms allows for example to quickly and reliably identify the following predictive segments:

  • the “customers at risk of churn” segment
  • the “high future value customers” segment
  • the “customers likely to buy such products” segment
  • the “customers in the process of being digitized” segment

These are strategic segments for a company, and yet impossible to identify manually: this is where predictive segmentation takes all its value.

Learn how DATACOOK automates predictive marketing segmentation, dynamically.

USE CASE: AUTOMOTIVE SECTOR

One of our customers, a car manufacturer that uses Datacook, has its own distribution network, a network of franchisees, a website, and call center platforms: each of these channels generates sales.

Manually, it is impossible to identify what are the criteria for re-purchasing vehicles, to predict which types of vehicles will be purchased after a first purchase, and which profiles buy which models. Datacook’s predictive segmentation algorithm allowed our client to answer all these questions, based on raw data collected internally and external data from open data.

It was thus able to identify patterns of repeat purchases of the same model of car on some of its vehicles. In other words, a customer segment that needs to be promoted to the latest version of the last vehicle purchased. 

In the same way, segments of customers who are particularly loyal to certain dealerships were identified in its network: the algorithm was thus able to restore the loyalty criteria and their weight, both on the customer side and on the dealership side, in order to understand the influencing criteria, and to act on what increases a customer’s loyalty.

“In other words, customer segmentation, when it is omnichannel, predictive and dynamic, is a performance gas pedal. It is thanks to it that you will be able to anticipate your customers’ expectations, and at the same time stay ahead of your competitors.


10. Dynamic predictive segmentation, the marketer’s asset

Unlike static segmentation, dynamic predictive segmentation updates segments in real time, and continuously injects them into CRM or campaign tools: the activated segments are therefore highly accurate at the time of activation.

Dynamic predictive segmentation takes into account the consumer’s real-time behavior AND their past behavior. This is fundamental to reconstruct customer behavior as closely as possible to reality.

  • Instant segmentation is carried out in real time, on hot data (data relating to the pages consulted during the visit to the site, the date of the visit, geolocation): for example, it segments Internet users in real time in order to push messages (pop-ups) or products (recommendations) to them in a relevant way.
  • Historical segmentation is based on cold data (data on age, gender, previous purchases, purchasing channels). It is based on information from the CRM, for example.

Dynamic segmentation integrates hot and cold data and continuously recalculates the segments. It thus restores the evolution of your customers’ behavior over time.

This type of segmentation requires an adapted solution, i.e. omnichannel (processing data from all channels), capable of processing both hot and cold data, and ensuring the continuous updating of segments, as customers regularly change segments.

For example, the financial sector uses DATACOOK for dynamic marketing segmentation of its customers.


Want to address your strategic segments in expert mode?

Learn how our customer segmentation platform helps you boost your ROI.