Hotel California + Data Science + Too Late Architectures
Had an interesting discussion with some colleagues this week. We wanted to translate some complex concepts — AI, Machine Learning, Deep Learning, Streaming Analytics, Predictive Analytics, API, Integration, Data Virtualization into a something that be communicated simply in a use case.
My colleague came up with something cool
- “DATA →”INFORMATION” →”KNOWLEDGE” — ->ACTION”
We talked about a real-world challenge that is present in quite a few verticals — “Customer Loyalty/UpSell/CrossSell”.
This issue is getting more important due to the migration to a “Subscription Economy”. Unlike the famous Hotel California song, customers can check out any time they like in a Subscription Model.
It is critical to ensure continuous customer satisfaction and success in this new model, both for classic customer retention as well as to monetize opportunities to cross-sell/upsell.
It is not a new problem and in the past companies have tried to address it via old-fashioned too-late Business Intelligence. I coined this term “Too Late Architectures” some time back for architectures that do solve the problem, but do it too late for the business.
Fortunately, now we have tools and platforms to build real-time software architectures that can address this problem in a much more efficient manner and provide monetization capabilities that the “too-late” architecture couldn’t.
DATA → Have access to all customer related information (static and in-motion) in real time. Is customer having service disruption issue right now? Is she about to call the customer service call center? Can I present a proactive message notifying of an issue to prevent dissatisfaction in the first place? Could I present a real-time offer based on current usage and product mix?
INFORMATION → What is my deep understanding of this particular customer? How well do I understand the current service issue? Do we have an ETA to fix any current service degradation? What specific offers are possible in the context of this customer, my products, and service level?
KNOWLEDGE → What are the patterns and relationships between service levels, customer behavior, and product purchasing tendencies? What types of proactive moves would be predicted to have the biggest impact? How does this service disruption compare to past outages, and can I learn from it to prevent in the future? What types of offers are most likely to be accepted and did I successfully improve my customer loyalty with this particular customer?
ACTIONABLE → This is a key capability, that allows you to react in time to an EVENT with an ACTION which is relevant, contextual and monetizable. Too-late architectures generally fail to proactively improve customer loyalty and instead up studying what went wrong.
In our example, I am able to make immediate, real-time moves: Proactively inform the customer about disruption before she calls the customer center. Offer her a discount on next month’s invoice.. Offer her a discounted offer to upgrade her package or add additional products and services that are relevant and timely to her, and her alone. Create an alternative, temporary way to keep her service at an acceptable level while the problem is solved.
Mandatory sales pitch — TIBCO platform is designed for this value chain (DATA, INFORMATION, KNOWLEDGE, ACTION). It allows you to leverage a full range of Analytics and Data Science capabilities (Real-time data acquisition and management, ML, AI, Deep Learning, Streaming, and Predictive/Prescriptive Analytics) to directly impact customer loyalty, lower costs, and increase revenues.
In addition to all these advanced analytics capabilities, the TIBCO platform provides a comprehensive Enterprise framework for Scalability, Automation, Model life cycle management, Governance and Security.
Check out a video https://www.youtube.com/watch?v=-bFKDJphxsA
