11.04.2019, Lesezeit: ~4min
In the four-part blog series on the topic of "Customer Behaviour Predictions", the Business Intelligence (BI) team presents an example project and explains how the team works together to develop a solution. In part 1 and 2, Fern Watson, who writes for the Data Science team, presents the technical solution. Sebastian Pisarski, expert for data visualisation, writes in the 3rd part how dashboards support the project. In this 4th and last part of the blog series Claudia Grünberger describes the marketing perspective. She explains here how we translate insights from consumer behaviour into actions.
In our BI example project, a grocery store is planning to simultaneously carry out promotions for meat and vegan products in its online shop. They have different advertising banners for the respective offers to choose from. In addition, there is a neutral advertising banner, which only shows a generic selection of articles. The food retailer is very interested in correctly assigning its customers to these product categories, because a targeted and precise allocation can lead to an increase in turnover. However, a wrong allocation of advertising banners can lead to people leaving the website without making a purchase.
Using a logistic regression model, we calculate the probability whether a customer is more likely to buy "meat" or "vegan" because of his or her behaviour on the store. Now the online marketing team of the grocery store is commissioned to specifically play out the corresponding advertising banners. If there was a probability of 75% or more that a customer prefers meat or vegan products, the appropriate advertising banner is displayed live online. If the probability was below this threshold, the customer sees the neutral advertising banner so as not to annoy him or her.
As a result, the grocer has to continuously monitor how customers' behaviour changed over time. In our example, the conversion rate after the launch of the targeted online campaign raises by an average of 8.2%. Clear proof that relevance can increase customer loyalty resulting in a positive impact on turnover. Key figures such as click-through rate, conversion rate or repurchase rate are also continuously monitored in order to optimise the algorithm, and thus the campaign, accordingly.
With the aim of maintaining a long-term positive success it is inevitable that the online marketing team visualises the corresponding key figures in a dashboard. Meaningful insights can, for example, be visualised with Tableau Software and presented in a way that is easy to understand. For instance, the Sankey Diagram shows that potatoes are an indicator for "meat" buyers in 65.20% of cases and for "vegan" buyers in 34.80% of cases. A result that surprised the food retailer very much and triggered a real light bulb moment in the Management Board in particular.
Customers of this grocery store have to register online when making a purchase. The food retailer can now collect purchase-relevant information about these registered customers, including which banner was shown and whether they clicked it, and address them in future via further campaigns. Examples of campaigns are: sending out relevant vouchers either by email or in the shopping basket, sending suitable product suggestions, and sending personalised eNewsletters.
The combination of a suitable channel and the ideal time to send messages plays a decisive role in reaching certain customers or customer segments. The use and popularity of online communication channels such as email, eNewsletters, SMS, social media, blogs, live chats or webinars is constantly increasing. These trends should be considered in every marketing communication strategy.
Cards & Systems has evolved from a classical technology company to an innovative full digital marketing agency. In the BI area, for example, we have developed a Business Intelligence engine that enables us to intelligently analyse and segment customer data. This allows us to support our clients in offering their end customers tailor-made offers through selected channels.
Business Intelligence Manager