Steven Noels / April 15, 2014
Consumer behavior data—such as buying trends, purchasing history, social media activity, etc.—is more abundant than ever….and companies of all types are taking note.
Financial services, telcos, retail, and publishing companies alike recognize that this data enables them to radically improve their marketing effectiveness, as campaign success is highly contingent upon timing and razor-sharp targeting. A multi-threaded analysis of customer behavioral patterns improves campaign efficiency, revenue and customer loyalty, but it also requires new prediction models and learning algorithms.
When it comes to challenges such as customer churn, risk or acquisition prediction, traditional tools are often limited to batch calculation of linear regression or classification models, and require significant manual tuning by data scientists. These systems—based on analysis and assumptions at a fixed moment-in-time—also have difficulty processing real-time (let alone constantly changing) data. As a result, they focus on past – or sometimes outdated – activity, providing little value to organizations looking to make business decisions based on accurate, up-to-date data.
This is where Customer Conversation Modeling (CCM) comes in. CCM – the next phase of data driven marketing – enables organizations to predict customer behavior before it happens. CCM focuses on three main areas and is based on the notion that people (customers) exhibit cyclical, multi-threaded behavior:
- Trend Detection – where sudden changes in behavior are more important than sustained behavior patterns
- Recognition of Cyclical Patterns – taking into account the time and location, and the depth/breadth of the historical conversation with the customer
- Multi-threaded Pattern Alignment Algorithms – tracking events across channels, aligning them in time and finding correlation between multi-channel behavior
For companies needing to provide greater predictive utility, it’s important to understand how to adopt CCM to help identify patterns, trends and event markers. At the same time, technological and architectural requirements are necessary to support this new model, starting from (big) data collection of behavioral information and transactional logs across all touchpoints. Behavioral data must be laid out on a normalized timescale in order to allow pattern recognition and alignment algorithms to work their magic.
Next, it becomes easy to use CCM to predict the Golden Path for every customer – the journey that provides the customer with the best possible experience, and gently nudge him to stay on that path with relevant service offers. Relevant in time, in context, relevant towards the customer preferences.
After all, Big Data isn’t just about deep analytics – it’s about providing actionable insights and drive operational marketing and customer service.
Grappling with how to keep up in this fast-paced, high customer expectations, data-driven world? Have you considered CCM? Tell us your story.