Maximize the Data Asset through an Integrated Stack of Analytical Tools

Rick van der Lans's picture
 By | februari 09, 2017
in analytical, analytical tools, analytics stack, big data, business insights, business intelligence (BI) architecture, business users, churn risk model, clustering, cooperating tools, data analytics, data discovery, data monetization, data products, data scientists, data standardization, data storage batch reporting, databases, face recognition analytics, forecasting. reporting tools, graph analytics, integrated stack, metadata specifications, reporting, sales and production figures, self-service analytics, sentiment analytics, super users, suppliers, transportation efficiency, Better BI, Business Analytics, Business Intelligence, Data Discovery, Data Quality, Data Visualization
februari 09, 2017

Sign up for webcast on Feb 15 with me on the next-gen analytics stack.

There was a time when organizations only used data for reporting to management and external parties, such as suppliers and legislators. The reports were commonly restricted to indicate what had happened within the organization. For example, reports would show sales and production figures, transportation efficiency, and so on.

But data is too valuable an asset to be used only for these straightforward forms of reporting. Better and more extensive use of data can lead to new business insights, which are needed in this highly competitive economy. Today, maximizing the data asset can make the difference between success and failure.

In those days, all the business intelligence (BI) systems were developed by professional IT specialists. They would design and develop the architecture, the databases, and all the reports. Users had to ask the specialists to develop and change reports for them. The reports would show totals, averages, and some simple forecasts.

With respect to the reporting tools, organizations tried to standardize on one tool; all the users would run the same tool. They thought that one tool would fit all. Now we know better.

Fast forward to today. The classic form of reporting was needed, is needed, and will always be needed, but we can do so much more with data. So many forms of analytics can be applied to data that can lead to new and valuable business insights. Clustering, forecasting, graph analytics, and many more techniques can be applied to show aspects of the organizations not yet identified.

All types of data can be analyzed nowadays, including text, audio and video. Everything from simple sentiment analytics to advanced face recognition analytics can be applied. Moreover, other groups of users besides IT specialists are developing reports and analyzing data. Business users, data scientists, suppliers, super users – they’re all potential developers of reports and data analytics to discover business insights.

This change of data usage has an impact on all aspects of BI. It influences the required expertise of our developers, it changes who develops the reports, it impacts the data storage technology used, it influences how data quality rules are enforced, and so on. And what definitely changes is, to meet all the reporting and analytical needs for this wide range of developers and data consumers, the one-tool-fits-all rule doesn’t hold true anymore. One tool will never be powerful enough. Organizations require a set of tools that together fulfill all the BI needs. These tools must support different groups of developers and support all forms of data usage, from the most mundane forms of batch reporting all the way up to the most cutting-edge forms of analytics.

The risk, however, is that organizations end up with a hotchpotch of tools, which leads to analytical islands  and on every island the wheel is reinvented again. It’s important that organizations deploy a set of cooperating tools that form an analytics stack, in which metadata specifications are shared.

For example, a specification for integrating two data sources developed for a simple report must be reusable by a data scientist developing a complex clustering model, and a churn risk model developed by a data scientist must be reusable in a more traditional report as well as in a self-service analytics environment. Additionally, besides being able to discover business insights, the analytics stack must allow an organization to operationalize and monetize these insights. In other words, the stack must be able to turn the insights into data products that provide actionable information without exposing their users to the underlying data stores or complex analytical models.

Data is too valuable an asset to be used for reporting only. An integrated stack of reporting and analytical tools that can be used by a wide range of developers is required to fully exploit all the data and that can drive an insights-driven organization.

I will be presenting a webcast on the next-gen analytics tech stack on Feb 15 - register here and learn how to figure out the classes of products to support different user types.