Thursday, November 8, 2018

Case Study: Creating a Power BI Portal on a Grand Scale

Key Challenges

   Consolidate multiple upstream data sources into a single, scalable cloud architecture.
   Automatically tabulate KPIs, allowing leadership to monitor data in nearly real-time.
   Feed partner-specific data to individual partners, allowing partners to track relevant KPIs.

Manually Compiling Reports

Our client, a large software supplier, needed our help to automate their partner sales attribution system. Partner sales were tracked manually, and this presented a host of problems. If our client wanted to identify how a product was performing with its partners, the sales team had to painstakingly compile the data from Excel worksheets. Compiling the information required a great deal of time. The data was collected, validated, made presentable, and only then shared with the leadership team. In short, the client would save a great deal of time if the process could be automated.

The client’s old system of manually tabulating the data inevitably led to errors. The errors undermined the integrity of the conclusions drawn from the data. Sometimes the errors led to misattributed partner rebates. For example, when multiple partners were involved in the sale of a product, the old reporting system sometimes indicated that multiple sales had been made. The reality, of course, was that the partners had all worked in conjunction to make one sale. Our implementation needed to ensure that partner rebates were properly scaled to the partner’s contribution to the sale.

The client also needed to be able to share partner-specific data with individual partners. The individual partner portal would allow partners to keep track of daily operations, monitor certifications, and compare rankings with other partners.

The client approached us to work on the project because of our long history of delivering timely results. We had worked with a key stakeholder in this project for the past ten years. We had grown with the stakeholder as he had climbed the ranks. When this project came to him, he came to us because he knew he could rely on us to get the job done.

Our Process

We began the project by identifying the appropriate architecture. We proposed three architecture options to the client and identified the various pros and cons. The client then independently reviewed the architectures and determined which one they thought would work best. As it turned out, we made the same design assumptions as the client, so we agreed on the architecture we were going to implement from the very beginning.

Next, we conducted several whiteboarding sessions with the client. During this process, we invited other teams (sales and UX, for example) to evaluate which Power BI views were most intuitive to users.

The whiteboarding process also allowed us to evaluate the client’s expectations. We discussed various client needs, designed mock-ups, returned with the mock-ups, and received client feedback. After the client approved our final mock-ups, we immediately began development.

The next step was to define the critical KPIs for the client’s leadership team and partners. We held meetings with key stakeholders over the course of two days to ensure that there was agreement on how the metrics were going to be measured. The first day, the stakeholders signed off on all but two metrics. The next day, we reached an understanding on how the final two metrics would be defined and measured.

After the job was fully defined, we entered our sprint cycles. We do not deliver a “finished” implementation without checking in. We submitted deliverables to the client as soon as they were finished. This continuous delivery approach guaranteed that we met expectations at each step. The approach also allowed our client’s leadership team to revise KPIs and project expectations as we progressed.

No project of this size is completed without significant challenges. Three weeks before the launch of the partner sales attribution system, we were notified that a critical component of our implementation was being updated. To resolve the issue, we worked closely with the component’s development team to update our implementation. We had one week to gather all the information we needed and just two weeks in production.

Results Delivered on Time, Once Again

We worked tirelessly to assure that we met the project launch deadline. The lead development operations engineer recalled how hectic the final three weeks of the project were: “We are lucky to have such motivated teams. They don’t get bogged down, and if something needs to be done, it gets done. They really make our work so much easier.”

It was a rewarding challenge to help our client implement such a massive project on such a short timeline. The Power BI portal accommodates over 50,000 daily users, consolidates data from over 50 sources, and delivers an array of nearly real-time information with less than five seconds of load time. Most importantly, we were able to once again deliver timely results.