The rise of the data-driven application for business intelligence
A couple of years ago, the term data-driven application began creeping into the business intelligence lexicon. Initially used by large vendors such as Microsoft to describe its newly launched Power BI visualization tool, data-driven apps have also emerged from startups. Young gun Reltio was one for the first startups to pitch its master data management (MDM) platform in the cloud for data-driven applications. Indeed, the term has now entered common parlance in BI and analytics circles. Yet it is not well understood.
The 451 Take
As has become clear to us when thinking about the wide variety of BI and data management providers that market themselves as purveyors of data-driven apps, this phrase is definitely in vogue right now. Like all buzzwords, however, it is likely to fall out of fashion and be replaced or augmented. But the idea behind data-driven decision management is a sound one, as the audience these offerings are intended to serve needs them, and implementing them can have a positive impact on a business' top and bottom line. While defining a data-driven application is hard enough, though, delivery can be an even greater challenge, calling for myriad foundation tools, as well as technology to deliver the intelligence component that is vital to differentiate them from dashboards.
What is a data-driven application?
Microsoft and Reltio's use of this nomenclature for two very different products serves to illustrate one issue with the term when it is employed in a business intelligence context. It is malleable and open to interpretation. However, there are a few defining characteristics to draw on for clarification.
First, as noted in our Total Data Analytics 2016 report, a data-driven application is one that actively influences evidence-based decision-making. Those skeptical of new marketing terminology might say that all BI tools should have this objective. We agree. Nonetheless, we believe this definition is useful to highlight enabling technologies for the creation of data-driven apps. Additionally, while BI tools might be deployed by a select group of business analysts and senior decision-makers, data-driven applications are designed to be used by the much larger group of business users who don't employ BI tools (see below).
Reltio's ability to support evidence-based decision-making via a graph-based cloud service is a prime example of an enabling platform for creating offerings of this nature. Data-integration and -quality technologies are other good examples of foundation tools used to create data-driven applications for BI use cases. The former is required to deliver meaningful insights on all of the data – historical, trend-based, streaming and event-based – appropriate to the data-driven app's role. The latter is necessary for accuracy and – along with MDM – for data governance and compliance, as well.
Data-driven apps can't be dumb or static
The 'intelligence' required is a second important aspect to consider. Microsoft Power BI is designed to bring that capability to a data-driven application via its visual analysis capabilities. Zoomdata is another example. The startup enables the exploration, visualization and analysis of data in various formats and state. Moreover, its offering has capabilities baked into it for making it easy to embed visual analysis so that the user experience is exactly what the data-driven app needs. Hadoop analytics specialist Arcadia Data is priming its latest release for data-driven applications by offering several capabilities for building, branding and sharing them, including a rewritten front end.
While visual analysis is important, there are other forms of intelligence that can be required for data-driven applications, depending on the use case. An offering of this nature may be required to make predictions, prescribe what to do next and – in some cases – take actions automatically. In these situations, predictive and prescriptive analytics that are driven by advanced analysis technologies such as machine-learning algorithms can also be required, as can natural language processing to glean insights from documents and other forms of text. FORMCEPT is an example of a vendor looking to address data-driven apps by providing many of these capabilities.
Audience is everything
Last but by no means least is the target market. The other defining characteristic of a data-driven application is the audience for which it is intended. True data-driven apps are designed for the general business user. This type of user has a low level of data literacy and scant analysis smarts because these skill sets aren't required in their day role. Business users want accurate, easy-to-understand insights that are germane to their role. They need an environment that can present insights from a variety of different sources in a meaningful, appealing way, with some suggested course of action. A basic example of this capability is offering recommendations to salespeople about which potential clients to prioritize.