Workday has made a continuous investment to be a data-driven business. In our early years most of this investment went to developing systems to gather alerts from our technology infrastructure, capture performance data, and track application access for security purposes. During those years we were probably better at gathering data than using it to optimize our operations.
That has changed. We’ve upped our maturity level with respect to data analysis, thanks to the efforts of development leaders who understood that to keep improving our service we needed to comb through the data. I thought it would be interesting to share a couple examples of our analysis efforts, highlight how a data-driven approach has helped our service delivery, and describe our ideas for continuing to gain more insights from our data.
I’ll start with an effort initiated by Jim Stratton, who heads our Technology Architecture Group. Upon joining Workday three years ago, Jim made a major improvement to our understanding of the performance experienced by our customers by looking broadly at our application log data held in a system we call Stats Warehouse (SWH).
Specifically, Jim analyzed performance for thousands of Workday transaction types across all of our customers for a period of two months. He categorized transactions into 50th percentile and 90th percentile performers. Jim was able to use the results to guide application developers to the areas of each product where improved performance would help the most. He started application service-owner meetings to regularly review and improve poorly behaving transactions.
These focused development efforts have helped Workday improve our response times even as the volume of requests continue to increase. Jim tells me he uses SWH every day and still is amazed at how the data leads him to areas for potential improvement.
Workday’s customers will continue to benefit from this increasingly data-driven approach to development.
Lynn Christensen manages all application development at Workday. A few years ago we set some aggressive goals for scaling Workday Financial Management, and to meet them Lynn and team turned to SWH analysis to augment what we were hearing from customers.
The analysis highlighted unexpected opportunities for tuning key areas in our financial management application that weren’t on the original roadmap. The data showed some tasks that we thought had a relatively minor impact on performance were having a much larger impact due to how frequently they were processed for some customers. The data also highlighted classes in our object model with unexpectedly high growth in numbers of instances over time.
Addressing these and other findings helped us successfully meet our performance benchmark goals and, more importantly, allowed us to get some large enterprise customers into production with Workday Financial Management without encountering performance problems they would have otherwise hit.
Lynn’s development teams have also implemented a program to monitor the transactions customers test in their previews of new feature releases and to look for any slow-running transactions. This program led us to find and fix significant performance degradation in a recent update before it ever made its way into production.
Find data-driven business leaders and give them the access, tools, and encouragement they need to really make the investment pay off.
We are excited about where this focus on data analysis can lead in the future. Our data sciences team is using machine learning to understand when we’ve loaded the minimal amount of data that customers need to run new updates, so we can restore their access to their tenants as quickly as possible. Application development is looking at ways to correlate key developer metrics, such as percent of test coverage and number of late check-ins, to actual performance results to understand how these metrics can predict future performance issues. Our customers will continue to benefit from this increasingly data-driven approach to the development process at Workday.
My advice for executives trying to encourage a data-driven approach is to realize that while you have to invest in systems to collect and store important data for analysis, that investment is not an end in itself. You should identify some specific things you want to improve and use data to guide how to address them. You also need to find data-driven business leaders and give them the access, tools, and encouragement they need to really make the investment pay off. You’ll find that generating a few success stories will be infectious and will help you advance from feeling like you’re drowning in data but not getting any smarter, to feeling like you’re truly using data to guide how you optimize your operations.