Artificial Intelligence in sales: While the use of digital channels in purchasing products continues to increase, conversion lags behind that of a personal advisor. What is the reason for this, and can the conversion level be improved by using new technologies?
Let’s imagine a classic situation in the personal consulting business: greeting with a handshake, small talk, and then getting down to business. With this picture in mind, it may not seem surprising that a consultant achieves the highest closing rates compared to other sales channels. Many years of personal experience paired with trained and empathetic behavior enables the consultant to react ad hoc to the customer’s every (not always necessarily articulated) wish.
Despite this obvious recipe for success, however, the potential is finite. The range is primarily exhausted, also thanks to digital support through lead generation tools, and the group of self-determined people is growing anyway, also thanks to digital possibilities. This trend is made very topical by the COVID-19 pandemic, even more so – who hasn’t cringed at the idea of a handshake to greet the personal advisor? The sales figures confirm this: eCommerce recorded a whopping 35 percent more business than the previous year. Whole product categories that used to be in the digital shadows are suddenly in greater demand online (e.g., groceries), and new groups of buyers are becoming active. Nevertheless, the conversion lags far behind the personal advisor, who is up to seven times more successful in closing.
The reasons for this are varied, but some can be traced back to the direct differences between digital and face-to-face sales. In contrast to the empathetic ad hoc reaction described above, digital application processes – especially at banks – usually follow a “one fits all” approach with a low degree of individualization, long adjustment cycles, and retrograde reporting, i.e., without continuous (and ideally automated) optimization, A/B tests excluded in the initiation.
If You Look Closely, You Will Recognize Patterns Of Success
At a time when artificial intelligence (AI) technologies have never been more accessible, individualization and, above all, automated optimization would be obvious solutions for approaching the success factors of personal sales. And in fact, there are already a lot of solutions with an AI background on the market, and CRM providers, in particular, are recognizing the potential of their products, which have always been geared towards the (data) analysis of user relationships. However, it is noticeable that such approaches usually work with personal user data and optimize them for expected target scenarios. But if the last few years have taught us anything, then it is
- that the processing of personal data, especially in the field of artificial intelligence, is viewed more and more critically and
- That customer behavior can change quickly, and yesterday’s predictions are no longer valid tomorrow – the effects of the COVID-19 pandemic serve as the best example here.
In conclusion, an ideal model would leave out personal data and works without previously defined target scenarios. It is a model that works purely based on process and behavioral data and calculates the plan itself—using the installment loan example. This model uses credit volume, term, or transaction channel criteria to find the path with the highest probability of success in the follow-up process.
Pattern Recognition By Artificial Intelligence
The following is conceivable: The AI finds typical patterns in the supplied process and behavioral data and classifies them accordingly. Possible success scenarios result from comparing comparable historical data that led to the conclusion. Based on the influencing factors, concrete recommendations for action to increase conversion can be derived if you now weigh the results. In order not to run the risk of chasing past-oriented patterns, the recommendations are constantly tested and optimized using evolutionary methods in test groups.
As with any AI, the quality of the results in this model also depends heavily on the amount of data available. Since, as already mentioned, personal data is not used, data pooling from comparable processes is conceivable to increase the volume. This means that data can be used in compliance with data protection regulations and across companies to build up collective knowledge for conversion optimization. This point, in particular, is crucial: If you combine the results from many processes, much deeper insights are gained than if each process were considered individually.