When people think of artificial intelligence, they often think of robots and futuristic concepts. So who would have thought that it celebrated its 65th birthday last year? AI is now indispensable and has established itself as a digital helper in our world. Alexa, Siri, vacuum robots or chatbots have long since become useful everyday and business aids. Against the backdrop of time savings, efficiency and optimized customer service, it is subject to continuous further development. 

The spin-off MARA Solutions is part of this development. The team has developed an AI-based analysis approach that enables companies to easily summarize and structure qualitative customer feedback without having to manually read thousands of customer reviews. 

Why?

Customer reviews are valuable assets for companies. They provide information about buyer satisfaction and show potential for improvement in products or services. Anyone who shops on the World Wide Web will find themselves clicking through other reviews before making a purchase - they have a significant influence on our buying decision.

That's why!

Conversely, for companies this means that all customer reviews must be considered equally. The problem: There are often several thousand reviews and it seems impossible to evaluate them all manually and efficiently. Above all, there are more every day.

MARA helps

"We ensure that qualitative customer feedback is quantified and evaluated in a structured way in the shortest possible time, without having to spend hours reading reviews," says Dr. Maximilian Lüders, co-founder of MARA Solutions, explaining the focus. The basis of the AI-based approach is the reviews on major e-commerce platforms, as this is where around 90 percent of the data for almost all product categories is collected. "We pick out the desired product category, copy the link and enter it into our system to develop a language model for it. The algorithm created automatically suggests the essential attributes in terms of product features and satisfaction, we just do a quality check. After that, the reviews can be evaluated with the help of the algorithm to make statements about the performance of individual brands or products," says Lüders, describing the procedure. In concrete terms, this means that in the first step, the algorithm reads out from the reviews what the topics are that are mentioned by shoppers in relation to product and performance attributes. In the second step, the algorithm recognizes whether they have rated the attribute positively, negatively or neutrally. Subsequently, these findings are summarized in a clear dashboard for the customer and supplemented by monthly reports with new reviews as well as competitors. Based on this, customers can derive their need for action for product management or marketing and check it after some time with newly added reviews. Automated interpretation of human emotions at the speed of sound - a revolution for product development!

THREE QUESTIONS TO DR. MAXIMILIAN LÜDERS, CO-FOUNDER OF MARA SOLUTIONS:

Who are your customers and what added value do you generate for them?

Dr. Maximilian Lüders: "We operate in the business-to-business sector, from small SMEs to large corporations. For small and medium-sized businesses, we are sometimes the first point of contact in the area of consumer insights. Many of them had no previous contact with market research, and we make this area accessible to them for the first time with our approach. But it is just as relevant for large corporations, who use it alongside conventional market research. The added value is the same for all customers: We help them improve products and target their marketing to better respond to customer needs."

What is the difference with market research and is there a competitive relationship?

Dr. Maximilian Lüders: "We see ourselves neither as a market research alternative nor as competition. Quite the opposite. We operate in the large area of consumer insights, within which market research is a long-established, very precise part. With some market research methods, one would arrive at the same results as we do, only in a different time, at a different price and not in continuity, because reviews continuously generate themselves, market research is done selectively. What is also frequently used within consumer insights is Natural Language Processing (NLP for short), simply put machine reading using artificial intelligence. Often, however, this does not go beyond 'better word counting'. But if I want to read the content of a review 'between the lines', an NLP tool is less helpful, because with a lawn mower, for example, the word grass is often read out as a so-called topic, but not certain attributes and their evaluation. Consequently, market research as well as NLP and our approach all have their raison d'être and can work independently or in combination with each other, depending on the customer's needs. However, we don't know of any offering in the market from data sourcing to dashboard creation like we do, so MARA Solutions is a useful addition for consumer insights."

What is the current status of MARA and what is the wish for the future?

Dr. Maximilian Lüders: "We are already out of the pilot phase since the beginning of 2020. We are selling the service and acquiring new customers. It's great to see that we can help companies work in a customer-centric way and that we can create an effectiveness experience with our idea. The positive feedback from customers encourages us to create a company from an initial idea, where great people participate and grow together with us. Currently, we are working on an even simpler and somewhat less structured way to summarize reviews for individual products and make their content even more quickly tangible. This would allow us to expand our customer group and use cases enormously, and the possibilities are almost unlimited - these summaries are also enormously valuable for individual consumers, for example. You could then have thousands of reviews summarized within minutes as to what the pros and cons are of a particular deodorant, lawn mower or lipstick. I think that's insanely fascinating."