The way businesses collect, analyze and use data is changing—from simply looking for and visualizing historical trends, to using advanced mathematical models to examine data in real-time and predict what might happen next.
Beyond that, this report from Tibco Software explains that the more switched-on companies are also starting to use technology to influence those outcomes, by combining data analytics with business process automation technology designed to enable companies to transform into truly data-driven businesses able to listen and react to what their customers tell them through day to day interactions.
Understandably, much of this development has been aimed at and led by data scientists, business analysts, and other specialists typically tasked with building data dashboards and applications for end users. However, with more data problems than specialists to solve them, vendors now provide self-service data analytics and automation toolkits that anyone can use to bridge the gap.
Such tools have been widely welcomed, however they are no substitute for specialist knowledge, and the issue remains of where, when, and how best to deploy them to create real business value.
Making the tools more accessible is only the first step…
Right offer, right time
Love them or loathe them, email and social media promotions have rapidly become a fact of modern life. Despite their ubiquity, however, they remain largely untargeted and poorly timed, often irritating rather than delighting customers and, in the process, jeopardizing future relationships. Make a new purchase online, for example, and you will set in motion an often very rigid and non-specific set of processes, starting with a rash of offers at best related to the product you’ve bought and at worse related only to the fact that you’ve made a purchase at all. Unless you opt out, you may also be added to a general mailing list and regularly advised of products, services, and offers that may or may not be relevant until you unsubscribe and, possibly, switch to a less persistent seller.
Data analytics can vastly improve this situation and make such marketing activity a great deal more relevant by considering not just what customers purchase, but their previous transaction history (where available) and what other customers have gone on to buy having made similar purchases.
Online retailers are well placed to do this, but it’s worth stressing that the processes are the same regardless of industry sector and whether the transactions happen online, over the phone, or in person.
It all starts with collecting as much data at the point of sale or customer interaction as possible. That information can then be matched with data from other sources in a process referred to as data wrangling, something that can be carried out using data analytics programs, which make it easy to link to multiple data sources whether held locally on-premises, in a remote data center, or in the cloud. There are tools that can enable you to analyze and visualize this combined dataset to make sense of the events being captured. In other words, to gain insight into customer behavior that can be used to make decisions on what to do next to influence the outcome of future interactions.
Understanding customer churn
Understanding why customers switch to another retailer or service company can be difficult and is often based on guesswork or gut instinct which, on closer examination of the data, will often be found incorrect, negating any steps taken to remediate the situation. Data analytics can be of great help here and, whilst in this case we’re looking at why customers choose to switch mobile phone providers, the principles are much the same across other sectors too.
On the face of it, there could be lots of reasons why mobile subscribers decide to switch providers, some of which can be addressed quite simply. The cost of calls, texts, or data downloads, for example, can lead to the common scenario of customers being offered better rates when they threaten to switch networks. In a lot of cases this incentive will stop them from switching, but it’s based on the assumption that cost is the main factor at work, which may not be the case.
The practice also costs money and might be missing an opportunity to address the issue for less, long before the customer takes steps to move away. To get a better understanding of what might be influencing churn, we would start by looking at the historical data held by our mobile network operator.
Visual analytics tools applying statistical and machine learning models can relate churn levels to a range of variables found in the data, such as contract length, total number of calls or texts, type of numbers called, call costs, and so on.
Although a major factor, cost is not the best predictor of churn. Instead, word of mouth is the best predictor – such as when customers receive calls from others who have already left. Such insight can now be used to target action, quite possibly before the customers (and subsequently their friends) even think about leaving as an option.