Digital strategy: how to approach personalization on your website?

    Personalization is one of the core elements of today's digital strategies. Learn how to use persona- or behavior-based personalization to create standout experiences for your website visitors.

    Subscribe to our blog

    For as long as I have been working at AMPLEXOR (that would be over ten years), personalization has been one of the top requested features. But only in recent times has it become feasible to embed meaningful personalization within projects that have to make do with real world budgets. 

    As web content management systems shifted more and more from pure content management and web publishing towards digital experience management (and thus marketing) solutions, most of them introduced tools for adding personalized content to websites and other digital communication channels. After all, one of the key principles in digital experience management is to put the visitor’s needs and wants in the center of your communication strategy in order to provide each visitor a meaningful customer journey. So in that regard, it is key to tailor the information you provide to the individual you are engaging with.

    In this blog post I am going to talk a little about several aspects of personalization based on our real world experiences.

    Personalization vs customization?

    If we’re going to talk about personalization it’s probably wise to define upfront what exactly it is, as the term is used rather loosely at times.

    The most important thing to realise is that there is a distinction between personalization and customization, even though these two terms are often used when talking about both techniques and tools.

    Personalization is based on implicit visitor data like pages visited, search queries performed, links clicked, social shares, likes, ratings, comments etc. A personalization engine will attempt to surprise and delight visitors with content that is very relevant to them, but that they most likely would not have found on their own accord. When it is working well, personalization will give your visitors an “Aha!” moment when they find something interesting or useful on your site that they would have never thought of themselves.

    Customization on the other hand is based on explicit preferences that are set by the visitors themselves. As a result, customization will, in most cases, not deliver any really surprising content to your visitors. Customization certainly has its usefulness, particularly in the context of intra- and extranets, portals, etc. Pure customization is just not very effective if you are trying to up- or cross-sell something to your visitors, as the customization engine will filter out any content that visitors think they are not interested in (even if they actually might be).

    As always, a real-life approach is not a matter of choosing one or the other; very often personalization can be augmented by some customization based on explicit user preferences.

    Personalization techniques: persona vs. individual based

    Broadly put, personalization solutions can be categorised as either persona based, or individual based.

    1. Persona based personalization

    Persona based is the more traditional approach and has been in use in marketing for decades, well before the advent of the web. With persona based personalization, the customer base is divided into groups with certain characteristics, and then individuals are attributed to a group based on the information they provide, either implicitly or explicitly.

    Persona based personalization

    This approach has been successful and is well understood, but there are some aspects that need to be considered very carefully in order for it to work. Personas have to be well defined and thought through. 

    In order to define personas, the business will need access to historical data and analyse and interpret it to arrive at meaningful persona definitions. If you need to assign an individual a persona, you need to have information on the person in questionSometimes this may be trivial, for example when a user has a login tied to a user in a CRM system. In that case, you may know their personal data and preferences set in their profile, their order history, a detailed account of their activity on the site etc. But sometimes you know very little, and that holds especially true for anonymous visitors.

    In light of privacy legislation, you may only know how they entered your site and their activity on the site within their current browsing session; you don’t know what they did last time - if there was even such a time - and what they’re doing now will once again be forgotten the next time they may visit your site. At these times when you know very little about an individual, it can be hard to assign them the relevant persona.

    In most off-the-shelf solutions, personas need to be defined and maintained manually. Also, managing what content is relevant for which personas has to be done manually. Manual intervention, as always, has a large cost and is prone to introducing errors.

    2. Individual based personalization

    When talking about individual based personalization, you no longer try to lump together individuals into predefined groups based on shared characteristics, behaviours or interests, but rather you try to target each individual as the unique person they really are. While this sounds great - and if done properly, it really can be - in reality, this type of personalization may be even harder to pull off.

    At the heart of these types of personalization engines lies an algorithm that will use behavioral data of the targeted individual and run it through a model to determine what content to show.

    So there is no team of content editors that will determine who gets to see what content exactly. The model in question could be a statistical model that takes into account behavioural data of other visitors (the so-called “wisdom of the crowd”). Well-known examples are the suggestions given by Amazon or Netflix.

    Individual based personalization

    In another approach, the model could also be based on inherent properties of the content to determine which pieces of match the visitors’ interests. This can be as simple as matching tags of visited pages with the tags on all other pages in your content database.

    Or it can be based on some sort of “DNA” fingerprint that is generated for each piece of content, so you can serve your individual visitor content that has a similar fingerprint to the ones of the content they already consumed. A good example of this is the music streaming service Pandora, that calculates a fingerprint for each song in their library, and serves up recommendations based on these fingerprints, rather than based on the music other people that listened to the songs you listened to, also listened to.

    • A big plus of the point above is that there is no need for a big editorial team to shape the content so it fits the personalization engine’s needs. On the other hand, the capture and analysis of all the data driving the personalization engine will of course also need some resources invested.
    • And again, as in the persona based approach, if you don’t know much about an individual, it is hard to determine which personalized content you need to show.


    Personalization is a major focus for marketers today. And it should be, as it is one of the ways toward a significant customer journey - every communication strategy should include some personalization in order to put the user first.

    There are two main types of personalization: based on groups of (semi-fictional) customers - or 'personas' - or based on the behavior of real individuals with the help of an algorithm. By using either one of those approaches and by implicitly tracking visitor behavior, such as links clicked or social shares, and using digital marketing tools efficiently, marketers can create a truly exciting digital experience for their visitors.

    Published on    Last updated on 11/12/2020

    #Digital Strategy, #Content Management

    About the author

    David Verdonck is Project Manager at Amplexor based in Belgium. David joined Amplexor in 2006 and has over 15 years of experience as a developer, analyst and project manager, specializing in Enterprise Content Manager and Scrum software development management.