*This post originally appeared on Tripolis (in Dutch).
The Tripolis Dialogues is a blog series that discusses modern marketing. How do today’s marketers deal with technological innovation as new privacy regulations go into effect? How do they cater to consumers who want relevant content but want to protect their anonymity at the same time? Unless’ Sander Nagtegaal was invited to participate in this series. Read on to hear his thoughts on “extreme personalization.”
The first .com bubble
I made my debut in the tech world as a programmer in 1999, during the first .com bubble. Since then, I’ve had all kinds of jobs. I was Chief Architect for Albumprinter, CTO at Peecho and later on, at Sumis. Then, I became CTO at MyTomorrows, a startup that tries to arrange access to drugs in development for patients with unmet medical needs.
The medical world deals with separate target groups, patients and doctors. It was difficult to speak to these audiences effectively at the same time. Unless we showed different content on the same website. I started thinking of a solution with my friend and co-founder Marcel Panse.
Our solution allowed us to target different audience groups on one channel, regardless of the code used on the website. We found that others were interested in this “serverless technology” and decided to turn it into a startup. The project captured Rockstart’s interest, so Marcel joined their coaching program. I followed later. The platform we developed set the foundation for Unless, which we founded in 2016.
A heavy sports injury
Before Unless started taking shape, I suffered a very bad sports injury. It was in the summer of 2012 during my Mixed Martial Arts training. I was hit by a stroke and had to learn to speak again. This was a difficult time, and not great for someone who was used to conducting sales conversations. I could no longer make my sales pitches appealing to different prospects. Which caused people to drop out.
My friends joked that I was like a static website, so I became intrigued with how websites communicate, which is mostly one-sided. That was the trigger behind Unless. A technology that helps you deliver relevant stories for every audience. It seemed easier to develop this than try to fix my own brain. With Unless, websites react more like a human being would. Your message is dynamically adapted depending on the visitor.
Extreme personalization: contextualization
Most marketers know about personalization. What’s even more interesting is not focusing on what you tell, but how you tell it. Basically sharing the same message, but packaged in a different way. By using pictures and testimonials that the receiver recognizes, for instance. You can do this (and more) by taking someone’s educational level into account, their mood and even the weather. That’s what contextualization offers - an extreme form of personalization. That’s our focus at Unless.
One of our customers sells toys. Their company has two clear target groups: parents and children. For the parents, you want to highlight that the toy is durable and made of wood. Children only want it to look cool. In short: the same product, different story.
Automatic audience segmentation
Unless lets you adjust your story depending on who is in front of the screen. To find out who that person is, we look at their location, time and day, and which campaign brought them to your website, among other things. This data helps you start segmenting your audience. For the most part, marketers have to come up with segments themselves. I’d prefer to see marketers define personas no longer based on gut feeling, but based on algorithms.
Our next step is to let websites segment target groups automatically. By recognizing patterns at a high scale (machine learning), Unless can classify visitors using large amounts of data. That data comes from previous website visitors and from our customers’ existing systems, such as CRM platforms, email marketing tools and remarketing solutions.
With this technology, you can predict what kind of person a new visitor is. What their expected attention span is, and the topics they find interesting. But also what their value as a customer will be (customer lifetime value), or what kind of costs your company will incur by serving them. Think of e-commerce companies that deal with costs from product returns, for instance.
Natural Language Processing
To garner your visitors’ attention, you must communicate with them in a language they understand. With words that appeal specifically to them. But understandably, you can’t expect to rewrite your website’s texts for each and every target group. With Unless, you will be able to automate that. This is possible because there are fixed rules. If you suspect that someone has a high lifetime value, but a short attention span, then you have be quick with your sales pitch. Having a short and concise message is key in this case.
With Natural Language Processing, it’s also possible to adapt a text depending on the educational level of the website visitor. You can go one step further, by customizing the sentiment evoked by the text depending on the visitor’s feeling. Suppose someone clicks on an ad. You can assume that the ad matches the person’s state of mind in that moment. That ad has a certain sentiment value, which you can determine with all sorts of algorithms. If you want the landing page to match that feeling, you must adapt that text to mirror the ad. So you can not only adapt a text based on keyword analysis, but also based on feeling. This type of symmetry leads to more engagement.
Every visitor gets to see a unique website when you combine all these things. We measure the success of each combination. On this basis, we separate the best performing options from the worst ones. This is called a multi-arm bandit test.
With the AI that we are currently working on, your website becomes a self-learning machine, which gets better and better at bringing in customers. A bit like Siri, Alexa and Google Assistant. This type of technology will change the internet completely. Websites will go from being primarily static, to conversational.
The filter bubble
When you think of personalization, you picture your Facebook timeline or think of product suggestions in webshops. The problem with this is that you create a filter bubble. We start thinking that the world works like Facebook presents it. Eventually, our worldview becomes widely different from others’. This can be dangerous, as it reduces our ability to communicate and understand those who are unlike us.
This issue can be addressed with large-scale automated contextualization. Messages that are difficult to understand for some people can be made more understandable, so that even if they don’t agree with the message they’ll give it a chance in the first place. Contextualization is therefore not only useful to stimulate purchases, it can bring about change in the world by improving online communication.
Go by the rules
According to privacy legislation, personalization falls under “profiling” and is permissible only if you have a very good reason for it, or if you have asked for consent. It’s important to understand when you need to ask for permission. The GDPR provides guidelines for this and sets the rules for how it should be done. Complying with the law is crucial, but you should also do your best to use personalization in an ethical way. When in doubt, follow the golden rule - ask yourself if you’d like to be treated that way. In theory, you can use Unless’ technology in the wrong way, but we do not allow it. We have established rules of conduct which determine when we should continue or terminate cooperation with customers.
I’ve found that technology has improved my life, or made it more interesting at least. But all new technologies create new problems while solving old ones. The amount of information we’re exposed to is so large that reality can get a bit foggy. If you don’t go along for the ride, you get pushed to the sidelines and you miss out on a lot. The same can be said for marketing technology. You can’t do without it if it brings added value, but you have to keep a critical eye.