Create, deploy, and maintain analytic applications that engage users and drive revenue. See a Logi demo

Embedded Analytics

9 Questions to Guide Data Architecture for Embedded Analytics

By Michelle Gardner | May 24, 2019
Share on LinkedIn Tweet about this on Twitter Share on Facebook

Data has gone from a byproduct of organizations to being crowned king of the modern world. Given the high demand for data as well as its complexity, data architecture has become increasingly important for organizations embarking on any data-driven project—including embedded analytics

A poorly conceived data strategy for your dashboards and reports can negatively impact the performance of your entire application. That’s why you need to carefully design a data architecture before you embed new analytics, dashboards, or business intelligence (BI) capabilities in your software.

>> Watch a Webinar on 3 Multi-Tenant Data Architecture Strategies <<

Inevitably, any modern data project will reach a point where performance and data complexity force you to re-evaluate the data architecture. When that happens, ask yourself these nine questions to inform your architecture plan:

#1. What are your data analytics goals?

Your data architecture will hinge on your goals. Are you generating reports to describe “what happened” (descriptive analytics), or giving people insight into what will happen based on the past (predictive analytics)? Or are you automating decisions based on patterns in data (prescriptive analytics)? Moving from descriptive to prescriptive analytics requires a progressively complex data architecture.

#2. Who is your end user?

Your data analytics goals will be tied closely to your end customer. Are your users simply trying to find information within the context of your application (for example, report writers gathering information from a portal)? Is your end user a data analyst who needs to understand performance and data structure? Is he a data scientist who wants to bring in datasets from different sources and manipulate them within your application? If you have multiple analytics users with different personas, you may need a data architecture strategy that accommodates scaling and progressive complexity.

#3. What types of content will be delivered and how?

Understanding what type of content will be delivered in your embedded analytics tool and how it will be delivered is an important part of your strategy. For example, a BI reporting tool that runs and emails reports has different performance characteristics and requirements than an interactive dashboard for an executive or a self-service BI user.

#4. What data structures do you have in place?

How your data is structured (e.g., array, file, record, table, tree) will determine how that data is stored and organized—and more importantly, how it can be accessed.

#5. What are your data latency expectations?

Data freshness is a metric that measures how quickly data updates are collected, processed, and made available in analytics reports. Do your end users demand access to real-time and near-time operational data? A well-designed data architecture can help you reduce data latency. Also consider the frequency of your data updates.

#6. Where are you in your project timeline?

Are you just starting out in your embedded analytics journey, or are you looking to optimize what you have in place? Are you looking for quick wins with plans to improve your analytics offering based on customer feedback, or do you have a longer analytics blueprint to integrate a workflow with more robust features?

#7. How do you plan to scale your current application infrastructure and platform?

Are you planning to scale up with new upgrades or scale out by linking to other resources? Growing data volumes will affect your performance, so data size matters. It is therefore important to design your data architecture with scalability in mind.

#8. What skills, tools, and budget does your team have?

Does your project team have the skillset to manage data issues? What tools are used to manage your data architecture? How do you monitor the different layers to detect performance issues? If you don’t have the skillset or tools, does your budget allow you to consult with embedded analytics vendors who can conduct performance reviews and provide advice and troubleshooting?

#9. How will you iterate over time?

Most embedded analytics projects are iterative, and will change as your users demand new capabilities and new technologies emerge. Your data architecture will evolve with the different iterations of your embedded dashboards and reports.

As you embark on your data journey, remember that the road to an ideal data architecture is long and rarely straight. Don’t expect to revamp your entire data architecture instantaneously. Your data journey will depend on your business priorities, your timeline, and the needs of your customers.

For more information, read our whitepaper: Toward a Modern Data Architecture for Embedded Analytics >

About the Author

Michelle Gardner is the Content Marketing Manager at Logi Analytics. She has over a decade of experience writing and editing content, with a specialty in software and technology.

Subscribe to the latest articles, videos, and webinars from Logi.