Use AI-Powered Behavioral Psychology in E-commerce

In a data-driven world, companies already have demographic, behavioral, and transactional data at their disposal, but they are often missing information regarding the motivation and personality of their shoppers.

By analyzing shoppers’ behavior (e.g., click-through, add-to-cart, and check out data) in response to specific persuasive messaging, it is possible to collect psychographic data that explains the psychological tendencies of individuals. This is what Crobox does. In essence, they bring AI-powered behavioral psychology to e-commerce.

I got to discuss this with Janelle de Weerd, the Marketing Manager at Crobox. With a background in tech, inbound marketing, and consumer psychology she has built and executed the digital marketing strategy. Coming on as the first dedicated marketeer for Crobox, she tried to establish thought leadership by bringing consumer and behavioral psychology to eCommerce.

Now, over to the interview!

Janelle de Weerd, Marketing Manager at Crobox

What is Your Company’s Background?

In 2014, co-founders Rodger Buyvoets, Leonard Wolters, and Sjoerd Mulder came together with backgrounds in eCommerce, artificial intelligence, and full-stack development. Initially, Crobox’s solution revolved around conversion optimization through the serving of persuasive triggers at each point of the buyer’s journey.

As the team grew, our solution evolved. The focus transitioned from purely impact and CRO focused, to unique psychographic intelligence centered. The team quickly realized that there was a gap in the market when it came to completing the customer profile puzzle. Companies already had demographic, behavioral, and transactional data at their disposal, but they were missing information regarding the motivation and personality of their shoppers.

With a strong base in psychology and persuasion, Crobox decided to invest further in understanding why shoppers behave the way they do. By analyzing shoppers’ behavior (e.g., click-through, add-to-cart, and check out data) in response to specific persuasive messaging, our solution collects psychographic data that explains the psychological tendencies of individuals.

Within three years, Crobox has grown to a team of 20 and is serving approximately 25 clients at the time of writing.

What Problem Does Crobox Solve?

Crobox helps eCommerce companies discover the why behind the buy with the delivery of unique psychographic data that boosts business performance.

What Does Crobox do, and why is it Important to Your Customers?

The Crobox platform offers a solution for eCommerce companies to gain better customer intelligence that makes an impact. By influencing shoppers to act at each step in the buyer’s journey, we provide tangible value to our clients.

This value comes in the form of three elements:


Crobox has a library of 100+ psychological principles that can be used online and offline to influence user behavior. These principles are adapted to various copy-variations, which are then tested with experiments at every step in the buyer’s journey.

When employing the experiments, we use The Journey of Influence (JoI) as our methodology. This method allows us to have a phased-rollout of experiments that are tailored to our client’s business objectives and platform opportunities.

For example, after signing up with Crobox, the first phase of the JoI targets decreasing choice overload. Choice overload is a barrier to (micro)conversion on the product listing page. So, to nudge users’ behavior in the desired direction, our software chooses specific products to feature persuasive messages.

Which, at this stage in the buyer’s journey, come in the form of product tags. These product tags can highlight features of the product with one to two words that appeal to shoppers’ subconscious.

For example, “Best Seller” or “Popular” relate to individuals’ tendency to look towards others behavior in times of uncertainty (Social Proof) and often triggers shoppers to click on the product. In a similar fashion, each stage of the buyer’s journey has a psychological barrier with a corresponding strategy and persuasive message/experiments to decrease the barrier.

croboxExample Product Listing Page Product Tags

After collecting enough data on the most effective principles (i.e., reaching appropriate power and significance), the next phase (or funnel step) will be unlocked with new corresponding strategies.

We can employ our technology on all steps in the buyer’s journey, from the homepage, product listing page, product display page, shopping cart, and confirmation page to off-platform channels and re-engagement campaigns.


Based on the findings discovered through our experimentation, we provide our clients with in-depth reports about their shoppers’ psychographic profiles. This gives our clients more clarity on the subconscious motivations of their shoppers and actionable insight into what persuasion tactics work best in triggering online behavior.


The result of using our technology has a significant business impact. Our clients benefit from increased (micro)conversion rates, average order values, (incremental) revenue, and traffic throughout the funnel.

While very few companies have similar software to ours, what makes Crobox stand out is its full-service approach. Each client is assigned a Behavioral Psychologist, who helps them to understand their target audience and results on a deeper level.

Also, our phased methodology remains flexible so it can cater to the business objectives of our clients. Meaning that clients can choose their journey after completing the foundational phases of the Journey of Influence. Which gives them the ability to put increased focus on high priority strategies or goals.

How Does Crobox use AI?

While other companies use A/B testing to test copy-variation, our AI does it for us, allowing us to test endless combinations of persuasive messages and products automatically. With the goal being to serve messages that are as relevant as possible for the shopper, our AI improves the effectiveness of each message that is served on a client’s platform.

When explaining Crobox, we say, “Derived from behavioral psychology, optimized by AI.” Our machine learning algorithms learn from the behavioral data of shoppers on a specific client’s platform. And, eventually, the AI will serve the best performing experiment/copy variation to users based on their location, device, and source.

On a deeper level, the Crobox AI has two primary functions. First, it finds the best products to highlight with persuasive triggers. Then, after uncovering the optimal products, it matches the best persuasive message to that product.

To address finding the best products for serving promotions, we have researched various recommender systems such as content, collaborative, and hybrid variations. At one stage, we experimented with context-aware recommendation systems, which take the continuously evolving context of a user or product into account.

However, we eventually found that we were better off creating our own variation that can handle our massive streaming architecture in real-time. At the time of writing, our solution is composed of several classifiers that “break-down” the matrices into simpler formats (e.g., tuples and triples), which can be asynchronously updated and is more suitable for extensive datasets.

To address the second function of matching the optimal promotion with the chosen product, we’ve invested even more time and effort over the course of the years. Our main objective in tackling this challenge was to find the balance between giving the system enough freedom to discover and ensuring the messages support the user in their path to purchase.

Ultimately, we’ve found that there is no blanket approach, meaning we have to use different strategies based on the client’s platform. Some of the strategies we’ve implemented include contextual multi-arm bandits, random forests, and NG Boost implementations. Next to these, we’ve developed our own variations that use mediation principles to select elements given the context of the user, product, and page.

These two functions often overlap, which presents a challenge in itself. Because the psychological experiments are always a combination of a product, a message presentation, and message copy, our AI has to consistently choose the best experiment based on the context of the user. This process of selection would be a very cumbersome and tedious job if done manually, making AI an essential element to our solution’s success.

In Your Mind, What is the Future of AI in Business and Marketing?

AI will continue to have a predominant role in business and marketing. People have grown accustomed to receiving seamless user experiences that are tailored to them on an individual level, whether or not they realize it. It’s the job of marketers to ensure that their users continue to receive relevant and resonating messages. And AI is currently the best way to achieve such relevance.

The one message fits all is a dying strategy. At Crobox, we believe in the one message fits best. And moving forward, there will be more companies using a similar approach to Crobox as more focus is put on using persuasion tactics to create better campaigns. AI will be the silver bullet in finding chaos in the mess of data and optimizing the most effective messages.

Do You Think Marketers Will be Replaced by AI Robots?

Of course, it’s hard to predict the future. However, if you consider that the computers used during the Apollo mission 50 years ago were a fraction as powerful as the smartphones we use today, It’s safe to say that anything is possible. But, at the same time, I believe that humans have certain characteristics that will always be needed in one shape or another.

Especially when it comes to the creative side of marketing, the storytelling side where authors have to build an emotional connection with their audience, I’m not convinced robots will reach that level of complexity. Unless we’re talking about singularity. In which case, the sentiment of the conversation may be slightly different.

When it comes down to it, I believe that as technology evolves, so does the human race. Our skills and competencies will develop, and our jobs as marketers will adjust accordingly. Call me an optimist, but I believe that AI is a tool to achieving better, more effective approaches.

Sure, it will shift the demand and focus of the labor force. But the markets will continue to adjust to technological development, just as they always have.

In this interview, Janelle de Weerd provided fascinating insights into how artificial intelligence can be used to collect psychographic data that explains the psychological tendencies of individuals for use in e-commerce. I’d say AI-powered e-commerce have a bright future!


I am an author, speaker and consultant in marketing automation and artificial intelligence.

Do you need help with marketing automation or AI-based solutions? Contact me and let’s discuss how I can help you!

Magnus Unemyr

Author, speaker and consultant in the aras of marketing automation, artificial intelligence, and the Internet-Of-Things. Contact me if you need help! Learn more.