#4 Blog – A Revolutionary Research Method enabled by AI

A Revolutionary Research Method Enabled by AI

A Revolutionary Research Method Enabled by AI


I love to go in more details here. So the basic idea is to have (from mythology point of view) the three-step approach. And again, the applications are across the whole buying funnel. Basically, what you can do everywhere is ask your target group a close- ended question.

Essentially, a rating question. Example: How well does this website make your visitors want to learn more or to book a demo or how much does this advertising make you want to book a demo – a rating question, right. With respect to rating, you can also have it in Amazon or whatever. And then, you ask “why” did you give this rating or “why” did you rate this way?

Thus, you collect quantitative and qualitative data and these couldn’t be elaborated. For instance, our survey bot is in real time, understanding what people are saying and then ask another question. So, you can get in very short time “in one minute” very elaborative input of what people really care about.

But the interesting point is, now this is still qualitative research, which has huge disadvantages, because you basically don’t know what it means towards behavior. Therefore, what you need to do next is – to do two more steps. And the first step is, measure with AI, because AI can help to ask another question.

So, you get really in-depth feedback in a very short amount of time. The second step is that, you need to quantify what people are saying. The quality of the feedback needs to be quantified in order to be able to put it into a predictive model.  So, we use here deep learning and we will talk about later how this exactly works, but it can be quantified (in a very granular code book) to tell us exactly what the person is speaking about.

And was this on scale? No matter how many people are answering, if it’s 50 or 500 or 5,000 or 5 million within milliseconds, you have an exact information what people are talking about. But now this is still not enough because you want to know which of those 50 things they typically say, are the killer things, which essentially drives behavior.

There is where a so-called causal, artificial intelligence, causal machine learning comes into play. I know this sounds a little bit scientific, but basically it measures how important it is and that’s all what counts. And what you see here is metrics ie.

X axis’ the importance, and on the y-axis is on how often the topics are mentioned and what you will see here is that they’re basically unrelated. The frequency of mentioning, has nothing to do with its importance, which is very interesting because if you ask your visitors or your clients, the question “why”, you would expect them that they will tell you why.

But in reality, they don’t know for most cases and they have no intention to really dig deeper. So, they give a true answer because you just ask an open question and they first answer something. What I mean is, if you ask your wife why did you take this jacket on today, she will tell you something because it’s not so important.

So that’s why there are some topics popping up, which are not so important. And you really need to find out which one are those. Therefore, this structure, well, this sole methodology, or which is basically nearly an automated system, can be applied on many, many different fields. And I would like to go through those fields now.

Free “How-to” Video on the Method


#2 Blog – What is AI not doing (today)

What is AI Not Doing (Today)?

What is AI not doing (today)? 


There are areas or things where it’s still not deployed. And that’s basically the whole area of strategic management. For example: what should I do so my people don’t churn? What should I do to refrain from making a bad advertising? What should I change on my website so my visitors convert? These are also abstract and high-level questions that AI and black box systems cannot or do not answer. 


Nowadays, on the other hand, they automate operative decisions. So basically, they look at what people do. For instance, the things they do are the input and the counter-reaction they do (for example, clicking a button) is the output and they can find the relationships in the process. But what they don’t address (simply because they don’t have the data available for that) is what people know, think and feel, when visiting your website


You just basically look on the data you gathered and try to do something from them, which is great. You need to use what you have. But there are limits. Limits of what your customers or your prospects think, feel, want, and know. And if only you could tap on this field, you could improve much better and you could answer those questions I raised earlier much, much better.

Free “How-to” Video on the Method


#1 Blog – What is AI in Marketing?

What is AI in Marketing?

What is AI? 

It’s very interesting and also because AI has been around for a long time. In here, this is me, and this is a book that I have published. This book is now over 20 years old to-date.  It’s in German, but let me translate it. You can imagine neural networks in marketing management and the applications described in this book are basically the same, as what you know today, because the main application is to predict and the prediction is basically used to automate processes.

So, if you can predict which person will respond to your targeting therefore, you can basically adjust your happening, or if a Google car can predict that there will be an obstacle when a person jumps on the street therefore, it can then push the brakes. So simply, AI is used for 99% in a black box way that it basically predicts things and we can automate certain decisions. 

So that’s what AI is doing. Nothing magical. It’s simply learning the unknown. Basically, the formulas. Unknown deterministic relationship between input and output from data and that’s what it does. secure our future.

Free “How-to” Video on the Method


Boost Conversion Rates with the Power of AI

Boost Conversion Rates with the Power of AI

More and more business models rely on the digital channel. No matter if you sell physical products, online fitness courses or you are collecting donations.

The challenge is well known. It takes an effort and money to get prospects onto your website. Sadly most leave the site shortly after. Those who stay often do not find convincing what they see. Those who manage to get into the checkout process often drop out for various reasons.

There are plenty of great tools out there to address these challenges. They analyze visitors’ traces. They help you experiment and test out different changes to get better.


Two Reasons Why Business Get Stuck With Conversion Optimization

Existing approaches have two significant limitations.

Limit #1: The data at hand has limited information.

You know where visitors are more likely to click and where they look at. You don’t know what they want, think, or feel.

Limit #2: You don’t know what you don’t know.

You can experiment only with components that you believe could have an impact. However, the number of things you can do are nearly unlimited.


We experienced it on our website. We had this wonderful landing page with a cloud-sunset video in the back. I loved it (because it was my idea 😉 ). Then we did this survey and asked two questions, “How likely will this page lead visitors to contact?” (Scale from 1 to 5) and  “Why did you give this rating?” (Open text feedback)

It turned out that 12%  mentioned the design to be outdated and that 12 % showed remarkably lower ratings. We reworked the design, A/B tested it and …. Voila, conversion nearly doubled.


This is How You Can Scale This

To scale this concept to your business, you need to solve these five challenges:

  1. Ask the right question, adopted to the page. The question is different on the landing page, on the product page, in the checkout, or when surveying trial or first customers.


  1. Automatically categorize text feedback in precision as if a human expert would do it. Only automation will allow you to analyze thousands of feedbacks at low costs.


  1. Match actual behavior data to the respondents’ data. Did this person drop, or did he buy? This data will help us later to predict the precise impact that certain topics will have.


  1. Find out how impactful the mentioned themes are. Correlation is not enough. It’s not enough to compare the average outcome KPI per the mentioned theme.Why? Because visitors mostly mention several themes. The only way around this is to apply advanced analytics (AI). This finds out which theme is the root cause of success. Then you will notice that the most often mentioned themes and reasons for ratings are mostly meaningless. AI is the key to insights.


  1. Make everything easily accessible in a simple, interactive and comprehensive dashboard (like this CX.AI demo dashboard)


If you do not have the time or resources to develop this process by yourself …

… there is good news for you

We did it for you. We apply it for great customers like Allianz, Audi, Microsoft, or Sonos – who love it.

This will not substitute what you do right now. It will boost its effectiveness by large. You will be able to detect powerful optimization angles you have never thought of.

For your visitors, it’s super easy and enjoyable. One rating question and one open text field. That’s it.

The validity is 4X higher than conventional approaches. The text categorization has double precision and driver analysis too.

Imagine this metaphor: You don’t feel well, and you need to see a doctor. One doctor has better machines that can make a better diagnosis (= text categorization). This same doctor also has more experience in this field and can better recommend treatments ( = driver analysis). Which doctor would you choose?

Now, you can use it too.

Just apply for a free consultation here.

Not ready yet? Drop me an email to


p.s. by the way, this is the newly designed website. Check it out