Who wouldn't want a crystal ball, that knew customer's intentions before they knew themselves?
The idea of using predictive analytics to understand and influence customer behavior has tantalized retail and hospitality industries.
It's not a new idea. Retailers, communication service providers, transportation and hospitality industries have used many predictive models to understand consumer behavior in the past. However, with changes in business models and consumer behavior, the predictive patterns change too.
On the one hand cloud, social and mobile have changed how consumers interact with businesses. On the other side technologies like Internet of Things, location, GPS and big data offered newer possibilities to learn more about the customer.
Some typical use cases for predictive analytics emerged over the years:
This involves monitoring prospect’s actions including their online evaluation and comparison of products and then sending them offers like coupons or using retargeting ads to incite them to make the purchase. Lately, app-based taxi services predict consumer buying patterns and send offers once they learn the customer’s travel plan.
We see examples of these at many e-commerce sites prompting “people who bought X also bought Y” messages.
Companies monitor interactions of their clients to predict attrition. Negative consumer sentiment in social media, browsing about issues on retailer’s online knowledge base, and repeat calls to contact center may indicate attrition. Service providers may use this information to triage the situation by offering free value-added services to prevent the consumer from switching.
This includes developing patterns to resolve customer issues quickly at the contact center, providing marketing and sales with guided recommendations on next-best-action to convert the prospect into a client, or uncover an upsell and cross-sell opportunity. Machine learning can be used to alert a sales call center that the best time to email the user is Friday evening when she is booking her travels for the following week.
Retailers and credit card companies monitor user purchase patterns and identify any malicious activities to warn consumers about any potential fraudulent transaction.
Still Coming Up Short on Consumer Satisfaction
Even with these well-established use cases and well-tried algorithms, companies still fall short in improving consumer satisfaction.
And with newer technology the situation is not getting any better, it is — for the most part — getting worse!
Businesses want to be agile, but the proliferation of new technologies and channels are creating more silos and more rigidity. Data remains fragmented across applications, warehouses, data lakes. Neither business operations nor business analytics have the complete information to make data-driven decisions.
The use cases above centered on the company, not on how to enhance the user experience. Companies developed them to meet departmental objectives, not necessarily to create pleasant consumer journeys.
The arrival of “the age of the customer” is causing companies to evaluate this challenge with a customer-centric lens.
We have more data than ever before. The number of channels has exploded, and we are capturing location, device, interaction and transaction data at a big data scale. Using predictive models and machine learning to be more customer-centric is the mantra of the day.
While the objective of any organizations remains to increase revenue, market share and reduce costs, the path to get there is changing — it's through a focus on customer needs and preferences. As customer experience becomes the new competitive advantage, companies use graph technologies, machine learning and predictive analytics to gain better customer understanding by:
Understanding consumer preference (product, place, channel) to offer right information and services at the right time and location – at the customer’s convenience
Grouping customers based on their social connections, locations and purchased products to understand their network by leveraging relationship graphs
Understanding consumer influence in their network, critical information to increase new product adoption
Identifying anonymous interactions and attributing to a particular user using identity resolution techniques
Grouping many individuals in a household based on various attributes and communication patterns.
How to Get Started with Customer-Centric Predictive Analytics
Companies seeking to deliver differentiated customer experiences must first build a reliable data foundation. Predictive analytics is only as good as the quality of data provisioned.
The following approach can help companies fruitfully leverage predictive analytics, gain deeper customer understanding and create the desired customer experience:
Learn and map how different segments of customers do business with you. Mapping customer journeys provides insight into their preferred channels, time and locations for various interactions. Use this information to predict the next-best-action to continue customer engagement.
Customers view the company as a single entity — not as separate marketing, sales and service departments. Therefore all departments need to share their data and serve customers based on insights from consolidated data.
There's no point in using predictive analytics to prevent attrition or to time marketing and sales campaigns if the sales, marketing, service and support data remain in silos. Keeping customer profile data, omnichannel interaction data, transactional data and analytics in segregated systems is ineffective.
It's not just the walls between systems that businesses need to remove. For predictive analytics to work, they need to bring down the functional and departmental walls between MDM groups, operations groups, analytics groups and data scientists. A shared data foundation allows each group to access clean, current and complete data and provides any data enrichment by one group to all in real-time.
Businesses need a continuous and well-defined program to measure data quality. Establishing data quality standards and monitoring data quality quotient in real-time makes predictive analytics reliable.
Leading organizations use machine learning and predictive analytics not only for enhancing customer experience but also to improve their data quality. Machine learning offers ways to match, merge and clean data and helps identify the data gaps, the poor quality data and recommend actions to improve the data.
Understanding consumer relationships and their influence in a community or household is critical information for retailers. Companies use machine learning to identify key relationships between users and help marketing teams with proper segmentation during a new product launch. Combining graph technology with Spark analytics provides ways to navigate relationships and to measure influence in the networks.
Relevant predictive analytics depends on having a reliable data foundation and uncovering relationships between people, products and places. Companies must ensure that data from all sources is blended and cleaned for proper understanding of customer journey, preferences and needs in order to deliver the most compelling experience throughout that journey.