Prior to cloud computing, taking on a data warehousing and business intelligence project often meant spending months...
acquiring hardware and software, implementing custom-designed data warehouses and meeting other business requirements. Moving to a cloud-based data analytics service allows you to spend more time on analysis and less time administering hardware and software.
Like other application areas, the cloud makes service delivery faster, less capital-intensive and more flexible. Both hardware and software are available on a pay-as-you-go basis. As you evaluate cloud analytics, or Analytics as a Service (AaaS) providers, consider how they implement the following key requirements: analysis and reporting, data loading and visualization.
Dashboards and ETL tools for cloud-based business analytics
Analysis and reporting combine to make the core of business analytics. And analysis and reporting covers a wide range of areas, from generating basic reports using internal system data to advanced data mining techniques that incorporate third-party data sources for an in-depth view of customer data.
Using Analytics as a Service can help enterprises obtain insights into customers and operations while reducing management overhead – though not eliminating it.
So, what do you need from a data analytics tool? There's no need to jump into the deep end of business analytics with advanced data mining in cases when all the enterprise needs is basic reporting. Instead, start with tools that allow you to deploy dashboards.
Dashboards are user interfaces that provide a view of multiple types of related data and allow users to search and drill down into pertinent data. When the marginal benefits of reporting begin to wane, enterprises should look into advanced analytics techniques. You don't need a Ph.D. in statistics to use these tools, but you will need a basic understanding of analytics techniques, such as classification, clustering, regression and text mining.
Additionally, extraction, transformation and load (ETL) tools and services can help streamline the process of moving data from source systems to your business analytics database. These tools provide higher-level functions than scripting languages that would be used to load data. For example, ETL tools can support complex workflows, track metadata and generate data-quality statistics. Rather than implement and maintain an ETL tool, you can use an ETL-as-a-Service model through some analytics providers.
Ensure that the ETL tool is compatible with the data source you require. Nearly all ETL tools work well with relational databases, XML sources and text files; however, mainframe data and enterprise applications, such as CRM and ERP systems, can be more challenging for some ETL tools.
Visualizations: Digging deeper into data analytics
Closely related to analysis and reporting is data visualization. Business data can be sliced and diced in many different ways. For example, you can look at sales based on customer demographics, geography or sales channels. Once you have a set of data, you'll want to identify problem areas, such as underperforming stores or untapped markets. Looking at rows and columns of numbers for these types of insights would be slow, at best.
Visualizations are designed to make contrasts in data jump out. Maps with color-coded pins representing sales by store, for example, can help an executive assess a large amount of sales data quickly.
When choosing which visualizations to adopt, consider the different types available and then deploy only the ones you need. Quality is more important than quantity when it comes to visualizations. It's also helpful to get feedback from end users; they will know better than anyone else which visualizations are intuitive and useful.
Using Analytics as a Service can help enterprises obtain insights into customers and operations while reducing management overhead -- though not eliminating it. Enterprise IT will be responsible for data stewardship, so look for services that have good metadata support. Business analytics metadata should include details about when datasets were loaded, names of source systems, definitions of data elements and quality-control metrics. This kind of data is especially important when you want to integrate data sets from different sources.
Data attribute names can easily be misconstrued, so keep detailed definitions for all attributes. This allows you to manipulate data sets for multiple reasons and can mitigate the risk of making decisions based on misrepresented data.
About the author
Dan Sullivan, M.Sc., is an author, systems architect and consultant with more than 20 years of IT experience. He has had engagements in advanced analytics, systems architecture, database design, enterprise security and business intelligence. He has worked in a broad range of industries, including financial services, manufacturing, pharmaceuticals, software development, government, retail and education. Dan has written extensively about topics that range from data warehousing, cloud computing and advanced analytics to security management, collaboration and text mining.