Currently, 1:1 personalization at scale is all the rage. Would you like to know how that works? Then, read on!
Personalization is a hot topic in the world of marketing. Its holy grail is automated 1:1 personalized context. It means that every single website visitor sees their own, unique version of your website - tailored to them specifically. Without a lot of work, and even if you have a lot of traffic.
It sounds like science fiction, right? Well, you could have it today!
Personalization goes way back. I clearly remember that slightly embarrassing moment of excitement after receiving an email from a big company, with a salutation containing my first name. Wow, they are thinking of me! After that early encounter with dynamic content insertion, product recommendations soon followed. Both are now commodity tools for marketers. But… there are way more exciting things now!
Currently, contextualization is all the rage. Contextualization does not adapt what you say, but rather how you package that message. Choice of words, images, examples and social proof are tweaked to appeal to the background of each unique visitor. This guarantees the largest attention span, the highest conversion rates, and generally the best user experience.
Successful 1:1 contextualization depends on three things:
I will explain how this works.
Before we dig into the mechanics of 1:1 contextualization, I should mention the use of “goals”. Like all marketing efforts, personalization only works if you have a clear view of your business objectives. These translate to “goals” in the dashboard of most personalization services.
For positive goals, you can think of visitors who buy your product, download a brochure or read an article all the way to the end. On the flip side, you may add people who returned a product, contacted the support department or canceled their subscription. All of those are useful to track as goals.
Your customer segmentation defines to which audiences your are catering on your website. An audience is not the same as a “user persona”. Although it’s a fictional character who reflects your customer type, a persona may be part of several audiences at the same time. Therefore, personalization services like Unless allow for multiple audiences that may be applicable to a visitor.
How does this work? Your visitor can become a member of an audience if certain conditions are met. These conditions can rely on:
Audience memberships can persist over time and therefore across web pages - even if you come back later. And this makes sense. For example, if you live in Amsterdam, that may be the case until you move somewhere else. Also, if you are above 40, you will remain above 40 for the rest of your life. And so on.
To provide for 1:1 personalization, it’s best to keep the scope of audiences relatively small, so people can fit into multiple audiences at the same time. Let’s consider this example. Say, I have three specific audiences:
How many different personas could we now personalize for? Well, assuming for simplicity’s sake that our audiences are not related and those people exist, this gives us 2 x 2 x 2 = 8 potential different permutations - or “kinds of people”. This adds up quickly. If you would use 20 of such audiences, there are over a million permutations. With 33, you could describe the entire world population.
Personally, I would fit in all three of our example audiences. So yeah, the guy in the lower right corner. That’s me.
So, we now know how to create many different audiences. Visitors may be a member of multiple of such audiences. Now, if you want to adapt your website funnel to provision all these different kinds of people, a similar approach is required for personalizing content.
Using Unless, I could change multiple elements on my generic homepage for the three different audiences of our example. I could replace the original headline with a pun from the eighties, add a different background picture showing a mountain trail and alter some text to reference a local barber shop in Amsterdam. The result is simple: for people like me, it’s instantly more recognizable.
Now, the important thing here is to realize that the personalizations fire at the same time, depending on your audience memberships. This means that all 8 different kinds of people that we recognized in our example will see a different result. If I am from Amsterdam, into sports, and only 33 years old, I will not see the pun from the eighties in the headline - but possibly a reference to Tamagotchi instead.
Using this approach quickly results in a very large number of potential page variations - in our simplified example it’s 2 to the power of n, where n is the number of audiences. Using many small audiences and many small personalizations, almost every visitor will see their own website. If you do this in a clever way, you will be able to reduce bounce rates and improve your conversion rates drastically.
Of course, in real life, many audiences are related or even mutually exclusive, limiting the number of potential page variations. Also, personalizations may be trying to adapt the same element at the same time. To prevent conflicts, Unless introduced a prioritization method for personalizations, in which only the most important personalization can change an element.
It works like this. When creating a personalized content snippet on a web page, this personalization must be prioritized with respect to the other personalization on that page. You can do that by dragging the personalization to the right position in the list. If, by any chance, personalizations would conflict, the higher ranked personalization will execute first and lock the changed element. This element can then no longer be changed by any lower ranked personalization.
While 1:1 contextualization at scale is already possible, there is more to come. At Unless, we are currently working on algorithms based on “supervised machine learning” that can predict the behaviour of unknown visitors, including the web pages that they are most likely to visit. This will improve the hit rate of your segmentation conditions drastically.
Secondly, “unsupervised machine learning” will be used to detect the most useful patterns in our data to automatically suggest the best audiences for your website. Using the funnel predictions, we will even be able to tell you which funnel to personalize for each of those audiences. Stay tuned for this - we expect to take it into production next year.
If you want to make the most out of your website, get the highest conversion rates and offer the best user experience to your visitors - wait no more. Book a demo at Unless.com to get an idea of how it would work for your use case. We will be happy to help you!