How do companies manage the overview of your customers to keep? How can companies be even closer to their own customers? How can sales and service employees always be up to date about current and future anomalies and potential?

These are exactly the questions that sales managers ask themselves every day! The answers to these questions can be answered with data analytics — or more specifically: (Predictive) customer analytics — be answered. To understand this in more detail, we have described five analytics use cases to make data-based decisions in sales in the future and find the appropriate answers to the questions above.

1) An overview with intelligent customer segments

Customers are often very different in terms of sales volume, order behavior or even the need for service. These differences must also be taken into account when analyzing customers. For this reason, it often makes no sense to speak of “the one customer group”, because anomalies and patterns are quickly generalized and the information derived from the data is not accurate.

Intelligent segmentation can be the key here to analyze customers more precisely and address them in a dedicated manner. Unlike traditional categorization, for example by region, turnover or product, intelligent segmentation includes a variety of relevant information (including customer attributes or customer features) and automatically creates the segments that best describe customer behavior in its entirety.

This is interesting information, e.g. for marketing and the derivation of marketing measures. However, it is much more important that this customer segmentation is the basis for AI-based analyses, such as migration forecasting or order forecasting.

2) Maintain an overview: abnormalities in customer behavior (anomaly detection)

Especially when there are many different customers or many different products in the portfolio, it is difficult and also very time-consuming to keep track of all customers in day-to-day business — even if they are segmented. Particularly among smaller or medium-sized customers who are not directly supported by one or more key account managers, abnormalities in their buying behavior are often noticed too late or not at all. This is where Anomaly Detection can help, which automatically and fully tracks all customers and proactively alerts sales to abnormalities in customer buying behavior.

These must not only be negative abnormalities, such as a reduction in sales for a specific customer, but can also be positive factors, such as the increase among a customer in a specific product category.

But how do you get these results (without much effort)? It doesn't always have to be AI... A more classic approach with descriptive analyses and built-in logic often works here. However, the aim is in any case to be aware of abnormalities early on — preferably on a daily basis — in order to be able to act accordingly.

3) Retain customers in the long term: churn prediction

Predicting customer churn is often one of the “classics” in the area of AI-based customer analytics. Especially when the overview of all customers and the buying behavior of these customers is confusing, potential customer migrations can be identified with the help of churn prediction even before they even happen.

This algorithm is trained using historical order, customer and contract data from internal systems (e.g. ERP, CRM, or service system) and often also external information. With the help of this historical data — and in particular information from customers who have already emigrated — patterns of customer behavior are identified before the migration. Current existing customers are then automatically checked for the identified patterns and thus a probability of migration is determined on a customer-specific basis.

It is particularly important not to leave sales staff alone with the pure probabilities, but also to name or suggest reasons and potential countermeasures. As a result, problems can be “tackled at the root” much more effectively, with the nice side effect that employee confidence in the analysis (or in the tool) and thus acceptance also increases.

4) Optimize the business: Analyze customer potential

In order to discover potential among individual customers, it is first important to identify the customers who still have potential and to calculate how high this potential could be. In order to make a statement about this, the so-called “share of wallet” approach is often used. This indicates how much a customer spends on the services or products offered compared to their total budget for these services or products. By comparing the turnover of similar customers (e.g. comparable industry, number of employees, turnover), sales potential can be identified. However, the relevant customer data is often not complete or correct in the systems — but enrichment with external company data can help.

Example: Customer A, who is very similar to customer B (both electrical engineering companies with 1,500 employees and an annual turnover of €200 million), currently purchases products for €500 thousand per year, whereas customer B sells 1.5 million € per year. In theory, customer A should still have a potential of €1 M, which could be exploited with appropriate products and measures.

And that brings us to the second category of customer potential analyses, because it is important to know which products can best exploit this potential. So-called recommender systems are used here. Recommender systems are also used by Netflix or Amazon. Systems that automatically generate suggestions based on similar customers (e.g. through intelligent segmentation) or even on the basis of suitable product and service combinations. Specifically: products that suit a specific customer in order to offer them to him or her with a high probability of success.

5) Expanding business: Data-based new customer acquisition

Another exciting use case in the area of customer analytics is looking away from existing customers and towards acquiring new customers. Although this statement is not entirely correct, because existing customers also play a very decisive role in data-based new customer acquisition.

Particularly profitable or high-turnover existing customers are identified and the characteristics and patterns of these “valuable existing customers” are derived. Specifically: What makes these customers special? Based on these acquired characteristics and patterns, very targeted queries can now be launched on external company databases, so that potential new customers are automatically suggested who are very similar to “valuable existing customers.”