Are You Ready for Data-Driven Apps?
The latest entry in TDWI's "Checklist Reports" series tackles a particularly pertinent question: What does the term "data-driven" actually mean?
- By Stephen Swoyer
- May 12, 2015
In the latest entry in TDWI's "Checklist Reports" series, research director for data management Philip Russom tackles a particularly pertinent question: What does the term "data-driven" actually mean? Russom contrasts the user experience (UX) of consumer-oriented data-driven apps such as Facebook and LinkedIn with that of the not-exactly-data-driven past.
His comparison is illustrative. "The success of consumer data-driven applications -- such as LinkedIn and Facebook -- has shown that frontline users can easily access, improve, analyze, and share their data in a seamless, unified environment," he writes. "The problem with traditional process-driven applications such as legacy CRM and ERP systems is that they place the burden of capturing analytical insight, making decisions, and taking action on the business user. To correct that problem and offer other benefits, modern enterprise data-driven applications [DDA] predict and prescribe what to do next with reliable data, relevant insights, and recommended actions."
According to Russom, an enterprise-grade data-driven app:
- Operates on diverse data from multichannel data sources, typically via data-as-a-service
- Provides near-universal data availability, from a single console or application interface; "The ideal DDA is a ... contextual application ... that customizes each view with accurate, timely, and relevant information based on the role and goal of the user"
- Achieves a "seamless" combination of operational and analytic capabilities
- Automatically generates master data, metadata, models, schema, and graphs; ideally, he argues, this should occur as a user searches, queries, or collects data
- Supports a "consumer-class" UX, a la Facebook or LinkedIn, and requires minimal training
Russom's new report, Bringing Modern Data-Driven Applications to the Enterprise, is a step-by-step primer for designing, building, managing, and improving enterprise data-driven apps.
The first priority, he points out, is to know what you're getting into and why you're getting into it. Identify a handful of compelling use cases for data-driven apps. Good candidates include affiliation management (a type of DDA app that tries to identify which customers "know" which other customers, as well as how they relate to one another), key account management (identifies real-world relationships between/among customers, accounts, products, vendors, etc.), or as a means to complement and support ongoing product development or a product launch.
"R&D and brand teams need a data-driven application for consolidating, sharing, and collaborating with product data drawn from multiple industry and enterprise sources[,] … [enabling] them to design, build, and launch products that align with customer requirements, competitive pressures, and market pricing," he explains. "The same DDA also provides reliable and complete product data for product hierarchies and groupings, manufacturing, and supply chain management."
Russom flags several use cases -- in life sciences, oil and gas, government, and other verticals -- that are industry-specific. The key takeaway, in all verticals, is that DDA apps must encapsulate a deep, domain-specific understanding of the business. "[W]hat makes DDAs unique is the encapsulation of deep understanding provided by internal stakeholders or third-party industry experts," he writes.
"Their knowledge of the data sets -- with functional and business processes fused into a DDA -- is ultimately what makes the application and the data relevant and effective."
Your second priority is to promote a data-driven work environment. To this end, a good DDA must:
- Deliver on Facebook- and LinkedIn-like levels of ease-of-use
- Promote user self-service and require minimal involvement from IT
- Permit a high degree of collaboration between and among users
- Provide access to up-to-date data, with an emphasis on right- or real-time delivery
- Make users more productive
Elsewhere, Russom writes, DDAs must consolidate operational and analytic functions into a single application experience, comply with and enforce governance policies as users work with data, and embed credible data management functionality. Above all, Russom urges, DDAs must not skimp on data management. "Analyses, reports, and decisions are only as good as the data they are fed. Don't just move data; improve data," he points out. "In other words, aggregating from multiple sources is an important first step. For the best analytics and business outcomes, however, improve data's structure and quality as well as its metadata and master data."
Similarly, DDAs must promote and enforce what Russom calls a "modern" approach to master data management (MDM). "MDM functions should be an integral part of an application's initial design. After all, most business applications are focused on managing a particular business entity -- or a short list of them -- such as customers, prospects, partners, products, supplies, assets, and financials, as well as their corresponding transactions and interactions."
Finally, DDAs are best implemented in the cloud, where the elasticity that's one of the cloud's strongest selling points can be exploited to greatest effect. "The virtualized server farms of a cloud -- regardless of cloud type -- elastically provision CPU and storage resources as data exploration, analytic sandboxes, analytic processing, and continuous data management demand them," Russom concludes. "As these workloads subside, the cloud automatically recoups platform resources and allocates them to other workloads. This assures linear scalability and high performance for data-driven processes without needing tedious capacity planning or overbuilding for peak loads."
Download a copy of Russom's report here. (Editor's note: A short registration is required for first-time downloaders.)