Imagine a bank that is about to launch a new product and offer it to existing customers. The problem is the bank can only communicate with their customers 3-4 times a year (otherwise, customers might qualify this communication as spam). Also, it needs to know which communication channels would be most effective for this purpose (and for every customer). So the goal is to offer the new product to existing customers having the highest chances to accept it – via the most effective communication channel.
In ACBaltica, we recently had a client who faced the same challenge. To address it, we used SAP predictive analytics tools (along with other SAP products, too). As a result, revenue in the retail segment grew by 30%, and the productivity of marketing campaigns grew by several times!Pavel Zhylko
Predictive analytics is a kind of advanced analytics that allow you to predict future events based on historical data. These tools usually rely on statistical modeling, data mining techniques, and machine learning. What data analytics and statisticians would do in weeks or even months, these tools can do in minutes (or days, for more sophisticated cases). And the best part is that the process is automated; the AI-based network keeps learning and gets better with more data it processes.
“SAP has several products that can perform predictive analytics,” says Pavel Zhylko. “Starting from a product called SAP Predictive Analytics to SAP Analytics Cloud (cloud-based analytics solution), SAP HANA (database platform), SAP Customer Experience (a suite of products for enhanced customer experience), SAP SuccessFactors (human capital management software), and more – they all have predictive analytics capabilities.”
To leverage the potential of predictive analytics tools, companies should first collect and manage all the relevant data – and define the signals that help to predict the future event. “Configuring and training of a neuro network for predictive analytics purposes might be tricky but let me explain it in simple terms, – says Pavel. – Our bank client had a lot of data about their customers (about 150 parameters) – so the goal was to identify which parameters influence customers’ decisions and score them. In some cases, a combination of the factors was also important (like salary + geography. To “train” the predictive analytics neuro network, we processed data of 10K customers that have already bought the product. Then, with another set of customer data, we checked if the neuro network was able to guess those who had already bought a product – and it figured out 94% of them! The next step was to process the data of all customers that hadn’t yet tried the new product through the trained neuro network and then – to contact those with a probability of purchase higher than 80%. Done!”
The beauty of predictive analytics tools is that it allows businesses to predict any events in any field where they have enough data: the performance of a marketing campaign, the probability of a candidate accepting the offer, the workload of a single employee, or a department, time to market of a new product and its ROI – and more. Based on customer behavior analysis, these tools allow companies to predict their future actions – and initially, lower churn and increase average checks and LTVs. For example, these SAP tools helped Raiffeisen Bank to increase the conversion rate of their marketing campaigns by 400% and decrease the number of “cold calls” by 80%. Turkish Finansebank used the same instruments to activate one-half of their sleeping clients within just four months and decrease the duration of their analytical projects from 6 months to just a few days.
For example, SAP tools helped Raiffeisen Bank to increase the conversion rate of their marketing campaigns by 400% and decrease the number of “cold calls” by 80%.
Theoretically, one could do some of these magical tricks with some open-source instruments (like Google Cloud Public Datasets) or even a sophisticated version of Excel. But if you have ever worked with huge amounts of data, you know the importance of visual and user-friendly instruments. “Some SAP products have predictive analytics models designed for specific industries. For example, SAP Hybris Commerce has a solution for a network of gas stations,” explains Pavel Zhylko. – “It allows suggesting the loyalty card users buy some products based on their previous purchase history. The suggestion is created on the fly while the customer is at the cashier to pay for the fuel.”
Another great thing about out-of-the-box products like SAP’s is that they allow creating an end-to-end solution: not just to analyze data and predict the event but also to determine the most effective communication channels, create and manage campaigns (like mass mailing, etc.), and then measure their performance.