Editor’s Note: This is the first blog in a two-part series that will explore the importance of continual data infusion, orchestration, and a closed-loop data cycle over the complete customer data lifecycle.
Ask a dozen marketers what it means to be “data-driven” and you will most likely receive a dozen different answers. Most will ruminate on the importance of customer data for creating personalized experiences that today’s always-on, always-connected customer craves. Fewer, though, will demonstrate a true understanding for how vital data is for breathing life into analytics and orchestration that are foundational for creating innovative customer experiences.
Continual data infusion ensures the long-term vitality of machine learning models, and also requires marketers to meticulously govern the underlying data orchestration to ensure that models are continually in sync with evolving business rules and outcomes. By grasping the true, full use case of customer data as the driving force for guiding and understanding a customer journey, marketers avoid common opportunity costs for underutilized data and avoid introducing friction into a customer journey.
Don’t Let Your Data Go Stale
A common misconception about AI in the customer engagement realm is that marketers can dump customer data into a machine learning model, sit back, and watch the magic happen. In reality, analytic models go stale over time. There is a cadence to developing models, deploying them, retraining them, continually testing them, and recognizing the precise moment when they have outlived their usefulness.
Continual data infusion helps dictate the data model cadence, answering questions such as when to redeploy or retrain. Also, new customer data could mean that an existing model might not be answering the right questions; different data, in other words, leads to different questions. A re-examination of models leads to a re-examination of the data, and vice-versa. New customers, products, channels, or ideas could all be the impetus for re-examining data or models to ensure the right questions are being answered.
Experimentation also plays a role; marketers want the freedom to test boundaries, and a broad range of questions requires a broad range of data to answer new questions. Multiple data models account for the innumerable intersections between a customer’s desire, marketers’ intent, and an organization’s constraints at any moment in time. Experimentation is about assigning certain data to a certain model to test, for example, the likelihood of a customer to buy based on a previous interaction.
What Is Data Orchestration?
Data orchestration is a process carried out by a piece of software that takes siloed data from multiple data storage locations, combines it, and makes it available to data analysis tools. A laser focus on data infusion as it relates to the vitality of data-driven models is half the battle for marketers for seizing the right moment of interaction with a customer. Data orchestration completes the picture, beyond orchestrating channel connectivity with a next-best action for a customer to include behind-the-scenes orchestration, with a system in place that ensures data flows where and when it’s needed.
Having the appropriate connectivity with automatic cleansing, merging, matching, and data transport, as well as automatic feeding into tools for model building, training, and assessment and into every task associated with orchestration is vital for marketers to achieve their personalization ambitions. An analogy can be made to building a luxury new home but skimping on the plumbing; a gleaming new data model will quickly become a dilapidated shack without a continual data pipeline that guarantees a proper flow from the data source to every orchestration touchpoint. Manually “carrying” data around via spreadsheets is a big culprit for missed data opportunities, introducing the possibility of data becoming stale, being put into the wrong campaign, lost or otherwise misused.
Dance with the Data That Brought You
An orchestration dance, if you will, is a symbiotic relationship between the marketer and the customer that is in constant motion from always-changing inbound and outbound touchpoints. Choreographed movements can be determined by marketer’s preference (I want to sell more of item X,) customer preference (I’m going to abandon a shopping cart to shop in-store), or by business rules or machine learning rules that select audiences, determine cadence, and have the dance’s next step planned out depending on the last step. The heart of orchestration entails measuring and acting on each of these preferences as they relate to one another in real time – which requires having the right data at the right time and the right place.
In recognizing that customer engagement – like the modern customer journey itself – is a continually ongoing process rather than a destination underscores the need to continually breathe new life into models or create new models in line with changing objectives, as well as to ensure that robust orchestration is in place to guarantee that customer data is not left behind.
In the next blog installment, I will dive deeper into data orchestration and focus on the importance of a closed-loop data cycle, including the continuous incorporation of feedback, retraining, and the customer’s voice.
What is Data Lineage and Why is it Important
Why Data Veracity is the Foundation for a Personalized Customer Experience
Data and the Empowered Customer