Transforming Microsoft’s audit practices with Microsoft Azure DevOps

May 25, 2021   |  

What if you could simplify your complex, manual, and ambiguous financial audit processes with automation and improved visibility?

With a little bit of creativity and valuable collaboration, the Microsoft Cloud and AI team did just that, developing a unique solution that helped the Microsoft Commerce Financial Services (CFS) team that it supports transform the way it runs its key SOX (Sarbanes-Oxley) financial audits.

The fix came from a desire to improve and automate the labor-intensive financial audit, which monitors internal controls and procedures relating to financial records. With over 900 controls to test annually, the SOX audit consumes 35,000 hours, generates roughly 1,000 Excel workbooks, and requires intense manual documentation.

Worst of all, there was a massive lack of visibility into what was happening behind all those manual processes.

But Michael Williams, an audit program manager in Microsoft Finance, knew the team had few resources to solve the problem. Poor communication tools, the inability to monitor follow-ups, confusion over completion status, and manual data entry created a work-management nightmare.

“We never had visibility to delays and issues,” Williams says. “When you’re managing hundreds of Excel spreadsheets, you can only see what people are willing to share with you.”

It was time for a change.

“Initially, we were talking about building automation and some items or tools from scratch,” Williams says. “We ultimately decided that may not be the best solution.”

Unable to develop an end-to-end answer on their own, Williams and the Finance team reached out to Microsoft Digital for help.

Enter Principal Software Engineering Manager Kishore Bondada and the Microsoft Cloud and AI team, including recent graduate-hire Thara Veeramachaneni.

“From the very beginning, Kishore wanted to understand the problem from an audit perspective,” Williams says. “What was the risk? How can we develop a solution that can meet the demands of business?”

The Cloud and AI team first proposed a web-based application that would support all of Finance’s SOX goals, but the two teams quickly recognized that building a portal would be a considerable expense.

“The cost of this solution was becoming a concern,” Bondada says. “It wasn’t a quick win for either team. To build a portal like this, with all the end-user experience requirements, would take time plus people.”

[Learn about how Microsoft Digital migrated critical finance systems to Azure. Understand how Microsoft uses anomaly detection and automation to transform royalty statements processing. Read about using AI and chatbots to simplify finance tools at Microsoft.]

How Azure DevOps became the solution

As luck would have it, one of Bondada’s previous Microsoft Digital experiences involved using Azure DevOps (ADO) to capture data from finance teams.

“We saw one of the previous implementations,” he says, “which prompted us to explore the idea, ‘Could we propose this solution to our business colleagues?’”

Realizing that an abstract conversation about DevOps might not connect with the Finance team, the Microsoft Cloud and AI team quickly created a form in ADO to show Williams how the idea might work.

“We didn’t know it at the time, but Auditing was already thinking about an Agile approach, so when they saw this template, they saw a connection,” Bondada says. “This could be a starting point.”

It was a roaring success—the test form affirmed that all of Finance’s needs would be met. A collaboration was now underway.

From the onset, the Cloud and AI team began researching the capabilities of ADO to determine which SOX work items could be ingested. Veeramachaneni worked closely with senior team members and Bondada in these early stages. Like most of the public, the university hire was unfamiliar with SOX, let alone what was involved, but this would later prove to be a benefit.

Bondada was pleased with Veeramachaneni’s early accomplishments.

“Initially, I was thinking we needed senior people full-time on this project,” Bondada says. “Integrating with ADO isn’t a straightforward process. But after the first few integrations, Veeramachaneni showed she was able to do it without supervision. The recent grad could also interact with the Finance team and get the work items done at the same time. She did more than just implement—she communicated across departments. I wasn’t expecting that from a newcomer.”

“I had a lot of mentorship,” says Veeramachaneni, referring to her relationship with the Cloud and AI team.

Having an intermediary who was new to the process was huge. Finance needed to learn the changes and how to use their new tools, and at the same time, they wanted to make sure certain functions matched their needs.

Veeramachaneni would play a critical role as translator and intermediary, asking ad-hoc questions while interacting with the Finance team and executing changes that matched their needs. Things that an engineer might overlook, such as authorization levels, could easily be communicated as feedback between Williams, Veeramachaneni, Bondada, and the rest of the Microsoft Cloud and AI team.

These regular discussions also allowed Finance to contribute to the final product. The Cloud and AI team started with over 50 entry fields, but Williams and the Microsoft Finance team helped to pare down that number and develop intuitive terms to make the interface usable to auditors.

Transformed audits paying off

Bondada and the Microsoft Digital team worked over the next two months with Williams to transform the SOX audit into an Agile model.

The improvements were immediately impactful.

Suddenly, Williams and the Finance team had the visibility they had always wanted. The solution met all of Finance’s review requirements, including security, accessibility, self-service reporting capabilities, and user experience. They could also customize it to include functionality and role-based permission controls.

“We could see the number of hours it takes to get through a control, the number of times we had to request information, the number of times we didn’t receive a timely answer, and now we have the data at our fingertips,” Williams says. “Now we can pull together a story and go to the groups we’re testing and say, ‘We see these things as hurdles from a control environment; you should probably be aware of them.’”

Williams can now spot bottlenecks in real time, meaning no more mad dashes at the eleventh hour. Progress can be viewed at overall and individual levels, down to specific control tests. Time-intensive and manual tasks are now automated. Summarized in hours, ADO significantly reduced the amount of time spent managing updates, documentation, and administrative work.

Williams is proud of the collaboration between the Microsoft Finance and Microsoft Cloud and AI teams.

“What we’ve proven is that ADO is a diverse set of tools that can be modified for a large-scale, dispersed, multi-tenant execution of products,” he says. “The teamwork that we’ve had with the Microsoft Digital team has been a true demonstration of One Microsoft and how we can break down barriers and really come together to solve a big problem.”

Bondada agrees, seeing the partnership as a successful use of Microsoft resources. “I didn’t ask for a lot of people to deliver a solution; that creates a lot of satisfaction.”

Leadership was thrilled, nominating the SOX solution for an internal award. Based on the success of this pilot, Agile Internal Audit is being tested out in other departments, including third-party oversight and investor relations.

And what did Veeramachaneni take away from the collaboration? “As a university hire, I was really excited,” she says. “I wasn’t expecting to have this level of impact on my first project with Microsoft.”

Learn about how Microsoft Digital migrated critical finance systems to Azure.

Understand how Microsoft uses anomaly detection and automation to transform royalty statements processing.

Read about using AI and chatbots to simplify finance tools at Microsoft.

Tags: , , , ,