cancel
Showing results for 
Search instead for 
Did you mean: 
cmdaniels
8 - Cloud Apps
8 - Cloud Apps

TreviPay’s Best Practice Journey

TreviPay introduced Sisense internally in 2017, shortly after they began constructing a data warehouse. At that time the data warehouse team was composed of a Database Engineer, Principal Architect, and Data Engineers who were also responsible for creating ElastiCubes and building data models in Sisense among many other things.

When Sisense was initially introduced the primary focus was to help the financial team consolidate information for financial reporting.  Then, there was a company drive to expand reporting for Customer Support and Client Accounts Receivable teams, and now nearly every internal department at TreviPay and many of our clientele access content in Sisense.

Let me shift gears and provide a bit more context into our setup so the remainder of the story makes sense.

TreviPay uses the Sisense on-prem solution. We are, and have always been, a single node cluster with two environments, development (dev) and production (prod). Our development environment connects to an instance of our data warehouse that is an exact replica of production with some limiters in place. We set up the replica environment to allow better testing and one-off ad-hoc support in our lower Sisense environment.

Now, back to the beginning, as demand for data grew and the designer count expanded things quickly spiraled out of control. Designers were only building content in production, designers using the same datasets would get vastly different results, and content was being shared with anyone who had access without any kind of review. Yikes.

Fast forward to 2021 where TreviPay formally introduces a Data & Analytics (D&A) team into the organization. Now there is a Data and Analytics manager, data engineers, architects, and analysts on the team with a Product Manager to guide them. However, the damage has already been done. This team is now faced with rebuilding the reputation of data integrity in the data warehouse.

The rebuild is set in motion.

This was obviously going to be a multi-step process. The first step was to determine how this newly founded D&A team was going to operate. The team set a path to consistently source and manage stakeholder requests and outlined a development process using agile methodology and tools.

Next, designers throughout the organization needed to be trained on best practices for building and distributing content. This focused heavily on driving designers to start building in the development environment instead of going straight into production and following a few simple design guidelines. For this, the D&A team put together training documentation using SharePoint for best practices they wanted other designers to follow.

Then, there needed to be consolidation and clean-up of fields available in multiple data sets within each model.  One of the most difficult challenges to overcome was helping designers understand the structure of the data models since they could not see them without higher licensing privileges. The D&A team chose to reconstruct a visual of the model using excel to depict the joins.

After that came the clean-up of old or unused dashboards in production. Using the REST APIs, we pulled together reports on dashboards including when they were last accessed, who owns them, who they are shared with, and which ElastiCube(s) they are using. We established a base set of rules to identify dashboards we wanted to eliminate and began purging.

The dashboards that remained were then divvied up by the owner or owning department. The desired outcome of this step was to eliminate any duplicative dashboards and review everything for accuracy. As you can imagine, this was a very long process.  The team worked with the dashboard owners to understand the purpose behind each of their remaining dashboards and determine whether they were still relevant. Each remaining dashboard underwent testing and review by the D&A team to ensure the dashboard was reporting the proper data as the designer had intended.

Fast forward to the end of 2024 when I am writing this blog, the D&A team has only grown by one individual throughout this time and process, adding another analyst to the team. Trust in the data available from the data warehouse has been restored and the team is operating more efficiently than ever. The D&A team has become the centralized reporting hub streamlining dashboards and reporting while supporting other analysts in their development and data questions. What seemed a daunting task has been achieved and TreviPay is better off for it.