ACADEMY / Personalization training blog
1 [Definition] What is segmentation?
Segmentation means dividing your market into identifiable and actionable subsets (segments). These share common characteristics in terms of needs, expectations or demand.
Customer segmentation is a key part of a marketing strategy: when you know your audience, you can put together a marketing mix that meets the exact needs of every visitor in that targeted segment. For those that don’t know, the marketing mix is the set of actions and strategies a company puts in place to promote its brand to the market. These actions traditionally cover four areas, known as the 4 Ps: Product, Price, Promotion and Place.
2 Which segmentation criteria to use
Segmentation involves creating homogenous groups made up of individuals with identifiable common characteristics. These might be place of residence, age, lifestyle or even how they behave on your website: these are what we call segmentation criteria.
The individuals within a same segment are supposed to have the same expectations and should react in a similar way to an offer, type of content or a specific message.
By cross-referencing different types of data, you can obtain a more in-depth analysis of your customers. Let’s review the most common segmentation criteria.
This type of segmentation is based on the geolocation of your visitors and is one of the simplest criteria. It enables you to target your marketing actions based on where your visitors are or on weather conditions.
For example, the international sports retailer Sportmaster has chosen to segment its audience according to their geolocation and local weather. It suggests products that meet the immediate needs of visitors. They see two different offers depending on their local temperature: a jacket rated to -10°; and another jacket rated to withstand temperatures of -30°.
Demographic segmentation is the most commonly used criteria, since it requires information that can be collected easily and that enable you to quickly target a potential market. These criteria include gender, age, nationality, education, profession, income or family situation.
For example, fashion websites commonly segment their audience by gender, i.e. displaying women’s or men’s clothing.
Psychographic segmentation focuses on the lifestyle of visitors: their interests, personalities, values, beliefs and opinions. To obtain this kind of information, you’ll usually need to have your visitors complete questionnaires or surveys.
For example, Club Med relies on psychographic criteria to optimize its users’ browsing journey by asking them to choose between “Sea” and “Mountain” when they visit the website for the first time. They are then segmented into two groups and if they return to the website over the next ten days they’ll be taken directly to the category they chose.
Behavioral segmentation relies on the way visitors interact with the website. Some data depends on their immediate online behavior (online) while other data depends on their past offline behavior (offline) when dealing with the brand.
- Online: time spent on the website, pages visited, point of exit, purchase opportunity (urgent or not), purchase attitude, brand loyalty (registering for newsletters), search engine and device used, traffic source, etc.
- Offline: number of visits, purchase history, date and amount of latest orders (RFM).
For example, to offer its visitors targeted deals, French supermarket chain Auchan relies on the user journeys on its website to segment its audience. Visitors who go to the “childcare”, “children”, “garden” and “furniture” pages or who add a product from these departments to their cart are targeted with deals relating to their areas of interest.
The different ypes of data to segment audience
Online, we distinguish between two types of data:
- hot data which is generated by the visit, such as behavioral criteria (browsing journey, history), contextual criteria (geolocation, weather) or technical criteria (browser/device used)
- cold data, or historical data, which you can find in your CRM system, such as demographic criteria (age, gender, socio-professional categories) or behavioral data from previous purchases (RFM).
To a large extent, this data overlaps with the segmentation criteria described earlier. Online, it’s this information that you will use as segmentation criteria.
3 Why segment your audience?
why is segmentation important for customers?
Traditional mass marketing enables you to find a compromise, satisfying the greatest number of people with the same offers. In comparison, segmentation enables you to zero in on the expectations of each customer. There is real demand for this, since 91% of consumers are more likely to buy from brands that offer them personalized experiences.
why is segmentation important for the brand?
Better knowledge of your customers and your market
Segmenting your market essentially analyzes who is part of that audience in detail and what characteristics you observe among your customers. It also enables you to understand which groups are the most loyal to the brand (or spend the most, are the least loyal, etc.) and, armed with that knowledge, better align your future marketing actions. This will ultimately enable you to offer your visitors an improved experience and to therefore retain your customers.
Better price optimization
It seems difficult, even impossible, to increase your prices market-wide overnight. However, by using a segmented approach you can identify the groups of people who are prepared to pay a little more for a specific uplift to your products or services. We’ll come back to this point in more detail when we discuss the applications of segmentation.
Greater value creation
With a segmented view of your market, you’ll be able to see a more significant ROI on your marketing actions than you would with a general approach. With a untargeted campaign that goes out to your whole market, the average success rate will be lower than if you successfully aim a suitable and differentiated campaign at several sub-groups. Essentially it is more effective to optimize segment by segment than it is to optimize for the entire market.
4 How do you segment your audience?
In terms of methodology, there are two ways to segment your audience: by choosing criteria "a priori" or by creating client types (the “post-hoc” method).
The "a priori" method
The "a priori" method, also known as rule-based segmentation, consists in manually dividing your audience into homogenous subsets according to pre-defined criteria. The criteria can be chosen based on the results of your data analysis or simply by using common sense.
For example a clothing brand could decide to segment its audience according to the gender and geolocation of its visitors. So, a woman living in London and visiting the website on 1st December would be presented with winter coats, whereas a man living in Spain who visited the website in the month of June would be shown swimming trunks.
Where do you start? To use this method, you already need to have an idea of the criteria that are relevant for segmenting your audience. The more detailed your market knowledge, the better chance you have of achieving efficient segmentation.
Limitations: This method may prove inaccurate, since a woman living in London and visiting the website in December could very well be looking to buy swimming trunks for her son who swims.
The "post-hoc "method
With the "post-hoc" method, also known as cluster-based segmentation, we don’t start with criteria or predetermined rules. Instead, we observe the similarities between visitors and then group them together according to those similarities. These are not fixed criteria (age, geolocation, interests, etc.) but rather a data set: buying behavior, visitor data, answers to surveys, etc. This approach reveals actual resemblances rather than hypothetical ones.
For example, instead of segmenting “men aged 45+ with an interest in cars” ("a priori"), we will segment “people who have bought X tires after being shown a specific promotion and who have stated that they own multiple vehicles”. This will then provide a “potential customer” segment for the automotive market that is more relevant than a segment built "a priori".
Where do you start? For this approach, you need to collect information about the visitors to your website (via data analysis, research, surveys, etc.) so that those with similar characteristics can be grouped together.
Who is this method for? The post-hoc approach particularly suits companies that have a limited knowledge of their market or are struggling to identify segments using "a priori" criteria.
Let's take the example of Rakuten PriceMinister, an online marketplace where individuals and professionals meet to buy and sell. When a user first visits the site, it has no way of determining if they are a seller or a buyer. No a priori criterion would enable it to clearly identify between these two segments. For this reason, PriceMinister uses the post-hoc method: thanks to Kameleoon’s predictive algorithms the brand can distinguish between these two segments and adapt its campaigns
5 Segmentation and targeting: what’s the difference?
Segmentation is when you divide your market into subsets. Of course, the segments identified won’t necessarily all be useful for your marketing strategy - if they are too restricted, for example, or if you aren’t able to reach them.
The trick is to target those segments where you’ll invest your marketing campaign budget: this is the targeting stage.
While targeting is an occasional activity (who is this marketing campaign aimed at?), segmentation is a more long-term approach (who are my current and future customers?).
6 How to prioritize which segments to target
Here are a few factors that will enable you to recognize those segments you should target as a priority:
The segment must be Relevant
The segment must be relevant, i.e. it must have a strategic value for your company and be made up of visitors with high added value. Moreover, to be relevant, segments must be different from each other, since they are not supposed to react in the same way to your planned marketing campaigns.
For example online tire retailer Allopneus has identified the “heavy drivers” segment. This segment represents only 10% of the traffic on its website, but it contributes 25% of the brand’s turnover: it is therefore a target with high added value.
The segment must be Measurable
The segment must be measurable and profitable, In other words you need a clear idea of the number of potential customers making up the segment, their purchasing power and their buying behavior. By analyzing these elements, you will be able to measure the profitability of the segment.
If you identify a segment made up of potential customers, but this only includes a small part of your audience, then it won’t be profitable.
The segment must be Accessible
The segment must be accessible - you have to be able to reach the potential within it through your marketing actions. Putting it another way, you must be able to communicate with your targets, whether by television, radio, social media or other channels, and also be capable of delivering the products they order to them.
Take the example of a brand wishing to appeal to a young segment – it needs to be present on channels such as Instagram and Twitter if it is to reach its customers authentically.
7 The applications of segmentation
Segment your audience, sure. But what happens next? How do you use this segmentation information in practice?
Personalize your advertising and communications
Design your advertising strategy based on the individuals that make up your market and their needs.
For example: If your audience is mainly made up of young children, you’ll want your adverts to be fun. The tone of your campaigns should be adapted to the segment you’re targeting. So, the better you know your target, the better you’ll be able to reach them.
With online retargeting, it is easy to reach again your former visitors with a specific message adapted to your audience.
Personalize the content on your website
Offer your visitors personalized content based on their interests, their geolocation or even their previous visit.
For example: On your website, you can switch the image on the homepage, create personalized banners or change content according to the segment that you have identified for each visitor.
Personalize browsing on your website
You can also adapt your visitors’ browsing journey based on their segment.
For example: You can highlight the categories most often viewed by visitors in that segment, reorganize menu sections according to their preferences or even personalize the results in the search bar.
Personalize your emails
Send personalized emails to your visitors after they leave your website.
For example: Remind your visitors that they haven’t completed their purchase or suggest products linked to their browsing or purchase history.
Personalize your deals
Suggest deals, such as discount vouchers or promotional codes, depending on the user’s relationship with your brand.
For example: Offer deals to new visitors to encourage them to become customers, to VIP customers to thank them for their loyalty or to undecided users to convince them to buy.
Design differentiated products
For example: Club Med identified two segments: families and people without children. It therefore designed two different clubs, tailored to the expectations and requirements of each segment.
Adapt your prices
For example: B2B pricing is frequently based on the size of the customer company and the needs it may have, with it often being set higher for bigger businesses.
8 Segmentation with predictive targeting
We have seen how segmentation is crucial to running effective marketing campaigns. However, brands still encounter many difficulties implementing segmentation. According to a study conducted by Forrester Consulting for simMachines,, 72% of marketers said that they had difficulty translating and analyzing data collected about their visitors, while 62% had issues creating content and personalized recommendations for each user.
This is where predictive targeting comes into play.
Predictive targeting involves using a machine learning algorithm to analyze website visitor data, enabling brands to identify which visitors belong to a particular target segment. By observing visitor behavior and the correlations between them, the algorithm learns how to predict the behavior of each visitor. Eventually, it determines – with increasing accuracy over time – which visitor belongs to which segment.
Through machine learning, the predictive approach offers superior segmentation, since it is capable of learning from the behavior of visitors to offer them ever more relevant experiences depending on their browsing journey. And it can do this even if it’s their first visit to the website.
why use predictive targeting?
Predictive targeting is used to identify segments that cannot be found using the a priori approach. It is essential when:
- the criteria that define the segment are vague, ambiguous or changeable (“I want to identify undecided visitors”, “I want to know which of my visitors will be influenced by scarcity effects”)
- the criteria that define the segment are too numerous and dissimilar, and impossible to model by hand.
In these situations, predictive targeting enables you to identify, with certainty, a greater number of visitors belonging to a segment that you’re seeking to target.
In this chart we can actually see that the predictive approach will identify almost four times more targets than the manual method.
what results to expect from predictive targeting?
Let's take the example of Allopneus, the French leader in online tire sales. Its goal is to target those customers considered to be “heavy drivers” to offer them a specific deal.
These heavy users are motorists who cover at least 24,000 km per year or who own several vehicles. They represent 10% of the website audience and 25% of its turnover. This is therefore a visitor segment with high added value.
With manual segmentation, it is not possible to determine which visitors belong to the “heavy driver” segment, unless they are already registered as customers. The predictive approach is therefore the ideal tool in this situation, allowing you to reach a target segment that is valuable but difficult to identify.
By cross-referencing hot data, such as browsing journey (brand, quality, budget, tire size), geolocation, source (SEO/SEM), time spent on each page and number of visits, with cold data, such as buyer profile (private individual, professional, type of vehicle owned), purchase history or form completion, the algorithm estimates which visitors have the highest probability of belonging to the “heavy driver” segment.
With this method, Allopneus can identify 48.1% more “heavy driver” customers than through manual segmentation, and has noted an increase of 15.7% in average cart value. Visitor segmentation is thus a key stage in crafting your marketing strategy, and one that requires special attention. It will help you zero in on whom to target with your actions and give you an edge over your competitors.