Technical

How to Build Data-Driven Analytics Applications

By Logi Analytics
Share on LinkedIn Tweet about this on Twitter Share on Facebook

Getting started with data driven applications

The newest wave of data analytics software providers presents a golden opportunity for full customization of business intelligence.  Not only are data analytics providers offering intuitive, robust, powerful visualization software, but they also are making it easier than ever to build or enhance our own data driven applications via embedded analytics.

Whereas custom applications that utilize powerful analytics would have required large manpower and capital investments in the past, they now can be built much more quickly and economically than ever before.  Embedded analytics negate the need for custom code and replace it with intuitive, context-specific visualizations.

These four main steps will get you started with your data driven application:

1. Build a core application from scratch or select an existing application to enhance

While the embedded analytics themselves require little to no custom coding, the complexity of the application that they are driving is entirely up to you.  You may choose to merely supplement an application that is already partially data-driven, such as your CRM or financial software, to develop a fully-custom core application, or to start in between with a partially off-the-shelf solution.  Either way, the first step is the same:  Determine the core needs and goals and select the technology that is best suited to satisfy them.

Note:  Because the embedded analytics that will drive the data driven application are so intuitive and powerful themselves, even a fully custom core application can be relatively barebones with little need for complicated code.  Particularly when establishing a minimum viable product, we merely need to ensure that the user can select the desired context.

Example

Acme, Inc. is creating a data driven application for its sales force.  The application will combine prospecting data from the CRM with sales transaction data from its in-house relational database.  For its prototype, Acme has decided to create a barebones JavaScript application that allows users to view information related to a particular customer, customer segment, sales branch, or salesperson.  To start, the application simply houses a blank page for each variant.

2. Select and build a corresponding data warehouse, if necessary

While many visual analytics tools can join data from disparate sources, any application will benefit from being driven by a single data warehouse that is optimized for the task.  If the application is being fed by millions or billions of records, this might mean loading data from all sources into a large-scale data warehouse like Redshift, BigQuery, or Hadoop.  Data driven applications based on smaller data sizes may be better suited to a relational SQL database.  In any case, select the data warehouse that best suites the application and create the necessary schemas.

In some cases, it may be desirable to utilize already-existing infrastructure rather than investing in an additional data warehouse.  Thankfully, we often can join multiple data silos within the visual analytics software with little to no lag.  The need for an intermediary data warehouse differs from case to case.  They key here is to select the technologies that are best-suited for a particular application.

Example

Acme, Inc. is a large, global organization conducting millions of transactions each year.  Although they have a relational database in place to house those transactions, they prefer not to slow down their production database by linking data driving applications to it.  Instead, they decided to create a separate data warehouse in Amazon Redshift, which will periodically update itself from both the CRM and transactional databases.  The visualizations in the application will access the data directly from Redshift.

3. Create reports and visualizations

Now that a data strategy is in place, we are ready to create the reports and visualizations that will be embedded into the application.  Our visual analytics software allows us to connect to one or many data sources that will drive the visualizations.  Whether we decided to create a single data warehouse or to connect separately to each disparate data source, we can now connect to the data through the visual analytics software.  We can then use an intuitive interface to build visualizations for our data driven applications, as well as to apply filters and other features to the visualizations.

Example

Because Acme, Inc. wants to provide its salespeople with embedded reports by customer, customer segment, sales branch, and salesperson, they need to create in-context reports for each possible variant.  They start by connecting to the Redshift database through the visualization software and creating an overall view that shows results across the entire company.  To create the in-context embedded reports, Acme simply needs to create a copy of the overall report and apply the necessary filters for each view.  These in-context reports will be the driver of each page that is viewed by the user.

4. Embed the reports and visualizations into the application

Finally, it is time to pull it all together.  The beauty of modern visual analytics software is that we no longer need to custom code each visualization into the destination application.  Instead, the embeddable code is generated by the software itself with each visualization that is created. To insert the visualization into our data driven applications, we simply need to copy the code from the analytics software and paste it into the core application.

Example

Remember that Acme, Inc. has already built a core JavaScript application.  They have also created a Redshift data warehouse as well as the context-specific reports that are to be embedded into the views.  As a final step, Acme copies the embeddable code that was generated by the visual analytics software, which also happens to be written in JavaScript.  For each view (by customer, customer segment, sales branch, or salesperson), Acme simply pastes the embeddable code into the existing code for the core application.  Acme has now successfully completed a minimum viable product that will power business intelligence for its entire sales staff.

Not only are data driven applications becoming more and more necessary towards retaining a competitive advantage, but they are also becoming easier and easier to build.  The overall complexity of the project will vary based on the nuances of the business and the technologies that are selected, but the core principals are the same.  To learn more about building data driven applications with embedded analytics, visit us here.

Originally published October 27, 2017; updated on March 19th, 2021

About the Author

Logi Analytics is the leader in embedded analytics. We help team put business intelligence at the core of their organizations and products.