Embedded Analytics

Data Strategies for Embedded Analytics

By Steve Murfitt
Share on LinkedIn Tweet about this on Twitter Share on Facebook

One of the most common questions application teams ask when it comes to embedding analytics is, what data strategies are important? Understanding what your customers are looking for and how their expectations have evolved is arguably the most challenging part of any analytics journey. It’s particularly relevant because analytics has changed from a “nice-to-have” capability to something end users expect within the applications they are using every day.

The important thing to remember is that you’re on a journey. Rather than trying to get to the endgame right away, you should work on how to get started. You can introduce things to your customers and then work with them on improving those things over time as you find out what’s working for them, where the gaps are between expectations and what you can deliver, and how to measure performance.

A Customer-Focused Approach to Embedded Analytics

Since your customers will expect embedded analytics with any application they use, you’re best off taking a customer-focused approach to developing those analytics. However, you’ll want to adjust your approach depending on whether you or your client will be owning your application’s content.

If you’ll be largely in control of the application – and the queries against it – you’ll likely have an easier time, since much of what’s happening through the application is under your control. You can create a simpler data architecture since your team will be managing that architecture.

If you’re expecting your customer to generate their own content, build their own queries, and interact with the data themselves, you need to be much more careful about how you present that data to them. What kind of experience will they have finding what they need? What calculations will they need to do that you can add into the application to help them?

Data Architecture is a Journey

How should customers and development teams address the challenges of data architecture in their embedded analytics? Treating the data architecture as a journey rather than trying to get the perfect solution right from the start is the best approach. To start, you need a clear understanding of and answer to three key questions:

  • What data structures do you have?
  • What data architecture are you using?
  • Is your data architecture designed for the types of analytic questions that your customers need to answer?

If your data architecture isn’t designed for your customers’ questions, then you’ll need to simplify how the data appears to your end users, and work on how to make that data architecture adapt for the performance and the requirements of those customers. Your goal should be to create a highly performant and very usable solution.

Managing an Evolving Data Architecture

Once you’ve established your data architecture, you’ll need a good strategy for making changes to that architecture when necessary. Just like with building your architecture, answering a few key questions early can help you in the long run:

  • Where are you starting from? This should be an easy question to answer, but especially relevant if your system is highly transactional, meaning it records your organization’s daily transactions. Examples of transactional databases are things like CRMs (Customer Relationship Management), HRMs (Human Resources Management), or ERPs (Enterprise Resource Planning).
  • What type of information does your customer want going forward? You’ll need to know if your customer wants to stick with information similar to what you have at your starting point or if you’ll be looking at much more aggregated data and trends over time. If it’s the latter, that data will generate different types of queries, and you’ll need to be prepared for them.
  • What will the load on your environment be? Understanding what kinds of information you’ll be dealing with helps you determine how much load will be on your environment, and whether your current platform is sufficient.

As your data architecture evolves, your team will need to make sure that your data strategy meets the requirements of your users and the performance requirements once people start using your embedded analytics. You can work towards improving your infrastructure to make that happen.

Key Takeaways

  • Analytics are no longer a “nice-to-have,” they’re a “must-have.”
  • Customers expect embedded analytics with any application. You need to take a customer-focused approach to creating those analytics.
  • For analytics, performance is key. When customers interact with the analytics in your application, they expect that all the activities are fast. This should be the primary driver of changes to your data architecture.

Originally published June 22, 2020; updated on August 17th, 2021

About the Author

Steve Murfitt is a Technical Account Manager at Logi Analytics. He has more than 20 years of experience in the analytics space, helping partners develop solutions to meet and exceed their customers' expectations with embedded analytics. He focuses on reviewing and advising on data and data structures to help present relevant information in a secure, usable, and performant manner.