“Data is the new oil” is probably the most cliched phrase in digital transformation. The pandemic has further proved this phrase to be true as businesses speed up their digital adoption to become more data driven. Data Lakes are gaining importance of being enablers of digital transformation as they address key challenges posed by data. Refer to the analytics spectrum in Figure 1, which shows the business value creation spectrum using data and advanced analytics. Predictive and Prescriptive Analytics capabilities are powered by advancements in Machine/Deep Learning technologies, enabled by Data Lakes.
Figure 1: Gartner Analytics Spectrum
Data Lakes can help leverage predictive and prescriptive analytics to provide incremental value to the following ten business applications.
Most organizations now generate a lot of data, much more than what their data processing and EDW platforms can handle. They want to do broader analytics and expand their BI capabilities to be more agile to the growing demands of their business. This could mean wider business adoption or more business insights from additional data sources. Data silos, scalability of databases and EDWs, inability to handle unstructured data sources pose an enormous challenge for BI modernization. Data-lakes provide a viable solution for businesses to process data at scale, adding unstructured and semi-structured data sources to provide better business insights.
Any medium to large sized contact center would have ten or more tools like call routing, IVR, Recording, CRM, WFM, QM, BI & reporting etc. to run effective operations. They all work in tandem to do just one job i.e. handling phone calls (mainly) and other inbound contact channels like chat, email, etc. However, these proprietary tools have trapped different aspects of the agent-customer interaction inside their data silos. This collective intelligence if aggregated from all data sources can provide insights not only to improve contact center efficiency but can give critical inputs to other business functions. Data Lakes can help reduce churn, improve retention, enhance products, increase compliance and help with agent productivity, voice of customer, NPS/CSAT and other metrics.
Recommendations have been very impactful, especially in E-commerce where recommendation engines suggest choices based on user behavior. Such recommendations require deep expertise for training and deployment of ML models. Data Lakes can process data sources capturing user behavior (click streams, past purchases), product inventory, demographics and feed them to models trained on historical data. Recommendation engines can create up-sell opportunities for front-liners (call center, retail store) by suggesting personalized options during customer interaction.
CX transformation is all about predicting and shaping customer journeys from an outside-in perspective. Companies store data from a single customer journey in disparate databases under control of different organizations. For example, healthcare ecosystem stores patient data across service providers and payor databases. As a result, the patient care continuum is fragmented. Data Lakes can help ingest, store and process CX data sources at a single place to get journey insights to improve overall CX.
Digitization has many definitions. In a very literal sense, it means converting information from analog to a digital format for further processing using digital technologies. In a broader sense digitization has been used to define modernization of processes, customer/employee engagement, business models by using digital technologies like IOT, AI/ML, Blockchain, AR/VR, 3D printing, etc.
More and more aspects of our lives and work are getting digitized with the use of smart endpoints like PCs, mobile phones, wearables, sensors, etc. Data Lakes can power your digitization workflows.
Data Lake in its very basic form enables storing structured and unstructured data. This is of no business use unless we derive some value out of it. At the very minimum, a data lake can provide an enterprise search capability. This means a data lake should be able to comb through the datasets (structured or unstructured) and build a catalog of its content. For e.g. in today’s world every company is generating a lot of natural language data in the form spoken and written forms in voice calls, meeting notes, company forms and documents, social media etc. A Data Lake can store and process this data and use pre-trained ML models to convert it into searchable structured data. Users can search through a user interface and find additional sources of business value.
Fraud is being committed daily across different industry verticals and government segments like finance, healthcare, social services, taxation etc. Descriptive analytics can process historical data from multiple sources to find clusters of fraudulent patterns, further used by predictive analytics to predict occurrence of fraud. Data Lake houses all the building blocks of an advanced analytics system.
Companies use time series methods to forecast future business outcomes such as product demand, resource needs, or financial performance. These methods fall short, as they do not incorporate irregular trends and independent variables. ML models consider additional variables besides time series data. Data lakes help with processing of time series data sources at a single place with other independent data sources to increase accuracy of business forecasting.
Manual processing workflows are prone to human latencies and errors resulting in delays and loss in productivity. Process improvements require a mix of data, process, and people to achieve desired improvements. Data Lakes provide a centralized data repository and processing engine to build automation workflows. NLP technology can help digitize natural language information and use it for decision support to automate processes.
IOT is bridging the gap between the physical and digital world by digitizing the attributes of physical things into digital formats. For e.g. a car being driven on road has many physical attributes related to speed, fuel, battery level, tyre pressure etc. which need to be in order for a smooth drive. IOT sensors in a car digitize these attributes, later monitored and analyzed for things like predictive-maintenance and driver behavior. Data Lakes play a pivotal role to enable these IOT applications by storing and processing semi-structured data streams combined with other data sources.
Sukhbir Singh Sethi is a Digital Transformation architect based out of Bangalore, India. His passion includes bridging the divide between Business and Technology by demonstrating business value of digital technologies. He works with customers across multiple industries to help them maximize business benefits from technology enabled initiatives.