A modern data strategy for a connected world

Yingying Kang, Director of AI & Data Science, Assurant

Data is a valuable resource.

Information is the most valuable resource in the 21st century. The Economist claimed, “The world’s most valuable resource is no longer oil, but data.” However, like oil, data cannot be used unless it is refined. Data needs to be decomposed, analyzed, and converted into insights to retrieve value. Today, almost all mainstream industries and businesses are on various stages of digital transformation or data-driven processes to reshape how they communicate with customers, reassess the health of their finance and operations, and renovate their decision-making methodologies. The businesses are competing to obtain and digest as much information as possible. However, this empowers companies to retrieve more information and brings growing pains to businesses managing 10 or even 100 times more data volume. Businesses need an optimized Data Management strategy to efficiently deliver the right data to the right users at the right time while mitigating the costs and risks brought by overflown storage.

Technologies are crucial. We live in a connected world.

The people in the 21st century live in a more connected world. We do not have to rely on a single solution to gain value from our resources or satisfy our needs. We can order Uber to catch 5 AM flights or rent a house for vacation on Airbnb to avoid crowds in hotels. We can manage our assets through smart devices even when we are thousands of miles away. We can share a bad shopping experience with friends all over the world or recommend services we had great experiences with. Almost all industries are developing online-offline networks to provide consumers with connected living experiences. Industries and businesses are competing to invest in new technologies like Internet of Things (IoT), Artificial Intelligence (AI), and Machine Learning (ML) to proactively predict customer needs and provide a better traveling experience. Very soon, we will no longer need to spend days to plan our trips and struggle through airport check-in nightmares when we travel. Airlines, hotels, and Online Travel Agencies (OTAs) are endeavoring to provide the most relevant tickets to their frequent travelers. Airports continuously invest in renovating security check systems, optimizing luggage management, and local commute strategies. These innovations are achieved with biometrics, AI, and IoT technologies. Similarly, Banking and Financial services compete to provide various intelligent products to help customers manage their properties and hedge risks in extreme events. We are living in a more connected world than ever.

"An effective data management strategy raises operational efficiency and reliability by establishing one source of truth and providing access to that data source throughout the organization."

Challenges to Businesses

However, more data means more operational costs to businesses. It is a big challenge for businesses to build a Big Data management platform to support the open-ended search needs with efficiency, effectiveness, scalability, resilience, and robustness. Today, many online agencies and IoT suppliers are facing pressures from markets to provide real-time services and sub-second search capabilities through their billions of transactions, which is beyond the traditional data warehousing capabilities. CIOs and CTOs are increasingly relying on Data Science to analyze business operations, identify key data flows, and generate effective data management strategies before investing in data platform or infrastructure. Depends on the business needs, the companies can apply Fog Computing and Edge Computing to manage decentralized computing, or apply Virtual Data Mesh or Fabric strategies to allow flexible data accessibility to distributed data systems, or apply centralized data platform to enhance data efficiency and ease of usability. No matter which strategy, the purpose of an effective data platform is to provide reliable information and remove ambiguity and waste caused by duplicated data feeds or reports.

Data Operations Management

There are many strategies to optimize the data operational costs. We can optimize the data storage with hot data on fast storage, warm data on slower storage, and cold data on the “deep but cheap” storage tier. Various storage technologies are developed to support business needs including NAS, HDD, SSD, memory fusion, data lake, and cloud resources. ML can analyze the patterns of data usage, which guides the caching strategies and retention policies to optimize the tradeoffs between storage costs and application performance. It also guides data transmission or data streaming strategies. Today the mainstream voices appeal businesses to move operations into cloud or hybrid cloud instead of maintaining multiple data centers on premises. Cloud computing technologies make auto-scaled and auto-managed computing capacity become possible. This enables businesses to focus on marketing strategies and enhance products and services for customers less worries about maintaining IT infrastructure. However, it also requires their technical teams to master cloud technologies and manage data environments properly and efficiently.

Data Availability and Transparency

Information asymmetry and isolation is another cause of operational inefficiency and customer failure. Consumers are often frustrated with long contracts, having no idea of the limitation of a service when they sign contracts with salespeople, and later find what they expected cannot be satisfied. Or they have to wait for next steps but have no clear expectation when issues can be resolved. Agents cannot give clear instructions to customers either because they are not offered enough training, or their internal processes are not transparent. An effective data management strategy raises operational efficiency and reliability by establishing one source of truth and providing access to that data source throughout the organization. Blockchain technologies enable businesses to build consistent and immutable transaction flows to exclude possible manipulations and duplicated efforts, which are typical of traditional business operations. IoT offers retailers better and real-time insights into their supply chain and store operations, allowing them to run more effectively. ML have been applied to automate operations. It also enhances process mining. By proactively detecting operational anomalies and identifying causalities, the issues can be addressed in advance.