“People have something in common: they are all different.” Robert Zend

The purchasing behavior and consumption patterns of Internet users are becoming more and more complex. They are more and more volatile, better informed and much quicker to say “no”. Worse, if your site does not offer a tailor-made experience, your visitors will go to learn and buy elsewhere …

Personalization is precisely what optimizes the experience of your users from who they are and what they want. This article is the first in a series in which we will discover the four steps to master to carry out your personalization strategy: Who? What? Or ? When?

Training & Co'm

Reading time: 8 minutes

This is obvious to marketers: personalization is managed by visitor segments. You cannot define user routes or offers and then wonder about the targets to which you will be able to present them. We adapt its content and offers to its segments, not the other way around!

What is segmentation?

To segment is to create a group of visitors who share the same characteristics (demographic, behavioral, contextual, etc.). The segmentation is based on the analysis – human or technological – of the visitor data. This is the first, essential step in a personalization process that will give consumers the feeling that they are considered unique. In the “physical world”, it is the relational supplement of the store seller that offers personalized service to put the consumer in the best purchasing mood. On the internet, it is the additional understanding in real time of the behavior of the visitor and their visit context that will allow them to offer the most simple, natural and successful shopping experience.

graphic - segmentation

1. Why segment?

Segmentation is a step that some marketers still neglect. In fact, some marketing teams still communicate undifferentiated across their entire database while hoping to achieve positive results.

Both theory and practice show that a visitor who benefits from his own experience has a much higher probability of conversion than the average. In application, this means that content suitable for each visitor. However, if your site attracts several hundreds of thousands of visitors each month, you will agree that it is impossible to create a unique experience for everyone, hence the interest in performing segmentation to identify visitors with behaviors, needs and with similar interests.

Segmentation is even already a prerequisite in many areas of activity. Being a habit, the visitor is gradually demanding. The fact is that all of the sites The most visited offer personalized experiences: the FNAC geolocates you, PriceMinister adapts its suggestions to your consumption, Renault contextualizes the promotion of its offers, etc. The visitor therefore becomes accustomed to an increasingly qualitative browsing experience which is now imposed on all advertisers to create a solid relationship between the visitor and his brand.

Personalization first needs to be aligned with the company’s customer relationship strategy: “on-site” personalization will not be expressed in the same way depending on objectives, typology and area of ​​activity of the advertiser. It is particularly important to understand the strategic targets and the CRM segmentation of the advertiser if it exists (RFM, PMG, attrition score, etc.).

2. The main segmentation criteria

Marketers must adapt to the needs of their customers according to criteria which can vary enormously, hence the importance of benefiting from a solution which makes it possible to constitute the complete DNA of its visitors to then target them according to any type of criteria. Data collection can come from two sources: on-site data and off-site data.

“On-site” data: Each visitor produces unique navigation data that you can retrieve. There are four types:

  1. Behavioral data : these are the actions performed by the visitor (visit time, pages visited, number of clicks, page scrolling speed, etc.). The targeting possibilities are endless since the behaviors that we are likely to study will vary from one sector to another. For some advertisers, the number or type of pages viewed will be a key indicator. For others, it will be the recency, frequency or time spent on a product page.
  1. Contextual data : these are the variables external to your site that can influence visitors. The visit context is often a powerful way to adapt your offer:
  • The time of the visit: day of the week, date range or time range;
  • Weather in real time or in the future: we know that navigation behavior and the propensity to consume are influenced by weather conditions, as reported in a study by PriceMinister-Rakuten.
  1. The origin of traffic : adapting its landing pages to visitors from SEO, sponsored links or social networks is a new word that is widely used on the web. Kameleoon can, for example, spot a keyword used on Google Adwords to communicate a message only to visitors from a specific acquisition campaign.
  1. Visitor characteristics : if you cannot identify all of the characteristics of each visitor, you can obtain advanced targeting of their profile and reduce uncertainty about them: old / new, language used, geolocation (by IP address), type of terminal used. It is even now possible to determine with a high probability certain characteristics or interests of a visitor depending on the sites he has previously visited: man / woman, interest in offers and products …

Off-site data: these are all customer and visitor data acquired prior to their visits. They are stored in a CRM solution (this is notably the case with socio-demographic and / or RFM data) or DMP (Data Management Platform). You can use them to carry out a first level of segmentation on the value of the customer according to its history or cycle of purchase (VIP, loyal, frequent, occasional, inactive …) or according to the type and amount of products purchased (from receipts in store for example).

Segment without DMP?

A DMP is essential to enrich customer knowledge, but not necessary to start in personalization. Segmentation possibilities are already immense based on the amount of data “On-site” that you can already capture from a personalization solution.

The use of a DMP will then enrich the range of data from which you can identify new segments. The combination of a DMP and a personalization solution can prove to be essential, for example, in a cross-channel dimension.

3. On what basis should you select your segments?

Regardless of the criteria that make it up, a visitor segment is only relevant if it can be exploited. There are three parameters to consider:

  • His value : The value of a segment will be high because it is strategic for your business or because it is made up of visitors with high added value (high purchase frequency, average basket, etc.) or because analysis shows that its potential for improving conversion gain is strong;
  • Its volume: does the identified segment represent a significant part of your audience? If you increase your conversion rate by 20% on a visitor segment which represents only 0.2% of your total audience, this ultimately only represents an increase of 0.04%;
  • Its feasibility : is it possible to collect data to identify this visitor segment? Difficult to identify, for example, an unregistered visitor who first visited your website via their smartphone before visiting your site again on the desktop.

4. Set up its segmentation

We differentiate two scenarios: either we already know its visitor segments, or we want to discover new ones.

  • You already know which segments you want to target: A personalization solution will allow you to “manually” build your segments by crossing the different qualification criteria: behavior of your visitors, context of the visit, source of traffic, type of terminal. A good solution will even provide you with “ready to use” segments with common purchasing characteristics (loyal, cart abandoners, etc.) or digital behavior (also called “buyer persona”: smart, stingy, dissatisfied, etc.) that can save you precious time.
  • You want to identify new segments: How do you know if you’ve already identified the most promising segments and those with the greatest margin for improvement in conversion? Your customization solution, if it includes an automatic segment discovery module, will meet this challenge. It crosses all the visitor criteria to identify the segments that outperform and those that underperform compared to the conversion average. In short, it shows you the optimization opportunities.

prediction - segmentation

Beyond that, you can also rely on predictive algorithms for situations where it is impossible to identify a segment by the multiplicity of the criteria which compose it or because the criteria composing this segment are by nature indefinable (for example those of “undecided” visitors). In this case, machine learning can take over; by learning about similar behaviors, it allows to push a personalized experience only to those likely to be in the right target. The role of the predictive engine will then be to choose the “best customers”, that is to say those who will ensure you the highest margin. In a way, you can delegate complexity and volume to the algorithms and focus on your business and your strategy.

You have defined your segments. You’re now ready to start creating personalized and optimized experiences. We will address the 2th stage of a personalization process, the “what?”, in our next article.