- September 11, 2019
- Posted by: admin
- Category: HPE
Getting Started with Embedded Analytics
All organizations should consider analytics a strategic investment. Not all organizations use the same strategy, however. Some organizations build their analytics platforms from the ground up to complement their core business solutions; others invest in prebuilt platforms that deliver the same (or better) functionality with more predictable cash flow. This buy-versus-build decision also applies to embedded analytics.
In this chapter, I discuss the first important decisions that you need to make before launching an embedded analytics solution. Throughout the chapter, I also provide guidance on how to decide between custom and prebuilt platforms so that you can choose the right embedded analytics solution for your needs.
Making the First Important Decisions
The decision to deploy embedded analytics isn’t a small one, because adding new analytic content and functions affects an organization’s
existing applications, services, devices, and websites. Before you embark on the embedded analytics journey, you must make several important decisions:
»»What’s the best way to deliver analytic content to users?
»»How can I ensure that the organization’s IT standards are maintained across all components and access points?
»»Should I develop the analytic functionality from scratch or license a third-party platform?
»»What are the key stakeholder evaluation criteria for a licensing scenario?
»»How can I ensure that embedded content is successful and engaging?
You should also consider how an embedded analytics initiative will affect the following:
Stakeholders: Although authority (and funding) for an embedded analytics initiative may reside with one business unit or executive, multiple stakeholders have equal importance: other executives, product and project managers, marketing and sales personnel, and developers. For more information, see “Stakeholder requirements” later in this chapter.
Existing applications and interfaces: Analytics initiatives are tricky to get just right, as are application and external interfaces. Embedded analytics is a product of both types of interfaces, so it’s doubly tricky to get just right. If you bring subpar embedded analytics into a highly regarded application or site, you may have user-satisfaction or adoption issues to deal with. To avoid such outcomes, seek feedback from all stakeholders, starting in the planning phase of the initiative.
Users: Embedded analytics users usually fall into two categories: external users (such as customers and partners) and internal users (who leverage the embedded analytics for a purpose such as gaining insight into a customer, patient, shipment, or transaction). Internal users tend to be separate from IT and other departments responsible for maintaining analytics.
Analytics platforms combine data handling and user interface (UI) components with features that vary in sophistication. When you evaluate analytics platforms, you have several technical considerations, such as depth and breadth of features, scalability, and extensibility.
Business considerations are easy to overlook but also require thorough investigation. You need to consider the following:
Total cost of ownership (TCO): TCO represents the total cost of a project, including direct and indirect costs. When it comes to in-house custom analytics solutions, it’s easy to underestimate direct costs and overlook indirect costs, such as third-party software components and internal long-term support costs.
Although a prebuilt analytics platform is never a truly turnkey solution, it’s easier to estimate TCO for a prebuilt analytics platform than for a custom-built platform. Compare how developer resources would be allocated on a prebuilt platform and on an in-house analytics solution. In
the latter case, the developers’ time would be allocated to developing the system rather than building and maintaining analytics on an ongoing basis.
Vendor expertise: Not far behind TCO in importance is the expertise of the original equipment manufacturer (OEM). Analytics-software companies generally have decades of experience in the field and could provide exactly the functions that users want, implemented in a way that supports current best practices in data visualization and other important areas of analytics.
Vendor support: Another factor in the buy-versus-build decision is vendor support. A vendor can help resolve data anomalies, performance issues, integration requirements, multitenancy deployment situations, and many other problems that can surface in an analytics solution.