{"id":5484,"date":"2020-07-09T11:48:21","date_gmt":"2020-07-09T18:48:21","guid":{"rendered":"https:\/\/www.microsoft.com\/insidetrack\/blog\/?p=5484"},"modified":"2023-08-07T15:34:24","modified_gmt":"2023-08-07T22:34:24","slug":"streamlining-microsofts-payment-processes-with-microsoft-power-automate","status":"publish","type":"post","link":"https:\/\/www.microsoft.com\/insidetrack\/blog\/streamlining-microsofts-payment-processes-with-microsoft-power-automate\/","title":{"rendered":"Streamlining Microsoft\u2019s payment processes with Microsoft Power Automate"},"content":{"rendered":"
\n
\n
<\/div>\n

This content has been archived, and while it was correct at time of publication, it may no longer be accurate or reflect the current situation at Microsoft.<\/p>\n<\/div>\n<\/div>\n

\"MicrosoftSometimes it\u2019s best to take matters into your own hands.<\/p>\n

Tanner Busby, Binu Surendranath, Russell Maw, and Korry Kelley aren\u2019t developers. But that didn\u2019t stop them from building a smart platform to automate a process that was stuck in the \u201990s.<\/p>\n

In 2019, Surendranath and Busby were part of Microsoft\u2019s global Accounts Payable team, while Maw and Kelley worked in Procurement. They were all familiar with how Microsoft handles requests to change supplier payment terms.<\/p>\n

Historically, when a finance professional needed to automate a process, they enlisted Microsoft Digital to build it for them. But thanks to Microsoft Power Automate<\/a>, they can now build solutions themselves.<\/p>\n

That keeps projects off Microsoft Digital’s plate and makes for quicker turnarounds. \u201cLike a lot of IT departments, Microsoft Digital\u2019s services are in high demand,\u201d says Busby, who is the service deliveries lead for the Americas.<\/p>\n

[Learn how Microsoft used Microsoft Azure to transform its cash flow business.<\/em><\/a>\u00a0Read about how Microsoft uses AI and chatbots to simplify finance tools at Microsoft.<\/em><\/a>\u00a0<\/em>Check out how Microsoft created efficiencies in finance with Microsoft Dynamics 365 and machine learning.<\/em><\/a>]<\/p>\n

Digitizing a manual process with Microsoft Power Automate<\/strong><\/p>\n

The team built the platform to solve a common problem: a deeply entrenched, heavily manual process was clogging up the gears.<\/p>\n

Each Microsoft supplier and partner enters into an agreement with Microsoft that stipulates, among other things, when payment is to be made to the supplier for goods received or services rendered. Sometimes, the supplier or partner requests that the payment be made in a way that deviates from the agreement.<\/p>\n

\u201cThose requests are common, and they don\u2019t just come from suppliers,\u201d says Maw, director of procurement services. \u201cThey might come from a Microsoft business group, procurement, or leadership.\u201d<\/p>\n

Almost all of the requests came in by email.<\/p>\n

\u201cMost of those requests went through my team,\u201d Maw says. \u201cWe would compare it to the original agreement and determine whether or not to approve the request.\u201d<\/p>\n

There was also no centralized approval authority.<\/p>\n

\u201cAnyone in the company could send a request to any Accounts Payable (AP) support alias globally,\u201d Busby says. \u201cAP would manage the ticket\u2014again manually\u2014and analyze it. They\u2019d send an email to Russell\u2019s team. Then his team would analyze it and make a decision, and send an email back to AP, who would pick it up and proceed. It was all terribly manual.\u201d<\/p>\n

As complicated manual processes tend to do, things got messy.<\/p>\n

\u201cThere were no criteria as to what to put in that email to request a payment term change,\u201d Busby says. \u201cFrom that point, it was back-and-forth emails trying to get the right data, the purchase order number, and many, many other criteria.\u201d<\/p>\n

The requests added up. \u201cI would get 40 to 50 of these requests a week, sometimes 25 a day,\u201d Maw says.<\/p>\n

The result was a drain on productivity, lack of visibility into the process, and discounts that weren\u2019t frequently applied. Automation was the obvious solution.<\/p>\n

Collecting data to iterate quickly<\/strong><\/p>\n

Surendranath, a group global process owner at Microsoft, built the first version with Busby\u2019s input in just five weeks. The initial release included minimal functionality.<\/p>\n

“We were just scratching the surface with the first version,\u201d Surendranath says. \u201cWe wanted it out in the world so we could collect feedback and iterate from there.\u201d<\/p>\n

He soon looped in Maw to create a standardized version of the process.<\/p>\n

\u201cIt wasn\u2019t just a matter of taking a manual process and then automating,\u201d Maw says. \u201cWe had to build the process. Then we built an app around that process.\u201d<\/p>\n

The proof of concept went live in July 2019. By December, the team had their first full quarter of data to analyze.<\/p>\n

\u201cWe got a sense of the volume, where the requests originated from, which business group, that sort of thing,\u201d Surendranath says.<\/p>\n

They also collected feedback from other requestors, their peers at Microsoft, and leadership. In many cases, the rapid iteration enabled by Microsoft Power Automate allowed the team to adapt the platform based on feedback and suggestions in near-real time.<\/p>\n

\u201cThere were many instances where users gave us suggestions or enhancements and the team finished the implementation of those suggestions before the meeting was over,\u201d Surendranath says.<\/p>\n

Korry Kelley, a senior data scientist, was brought in to flesh the platform out.<\/p>\n

\u201cKorry did a lot of the initial creation,\u201d Surendranath says. \u201cHe brought some best practices over from his experience with the Microsoft product team and drove our scalability efforts.\u201d<\/p>\n

To save time, Kelley reused some components from previous projects like the layout, skin, and data source.<\/p>\n

\u201cV1 was basically just an input form,\u201d Kelley says. \u201cWe were collecting data. There was no extensive reporting, it just gave us an indication of volume.\u201d<\/p>\n

In December 2019, the team got to work on V2, which was released on February 1, 2020. With plenty of data and user feedback to draw from, the platform began to look like an enterprise-ready product.<\/p>\n

\u201cV2 was a one-stop shop, a single window for submission, approval, tracking and configuration,\u201d Kelley says. \u201cWe also built out the operational management functionality and added triggered notification emails to stakeholders. We gave process admins the ability to add and remove finance controller approvers, and for approvers to delegate that responsibility when needed. It was a groundbreaking user experience in a line-of-business offering. You don\u2019t generally see that kind of experience without going to IT and having them build something from scratch.\u201d<\/p>\n

\"Diagram
The payment term change request process flow.<\/figcaption><\/figure>\n

In the new platform, a business stakeholder kicks off the process by submitting a new request for a payment term change in the app. Based on the decision engine and business rules configuration, this request gets routed automatically to the corresponding finance controller, who\u2019s notified by email and in the app. The finance controller reviews the request, asking for any additional details that they may need through the app. When the request is approved, Finance Operations executes the approved changes.<\/p>\n

Microsoft Power Automate<\/a> executes the entire workflow, reading and writing the supplier data to the Azure SQL<\/a> backend and transactional data to Microsoft SharePoint<\/a>, and sending emails to any relevant parties that need to take action. Stakeholders can view a snapshot of queued requests in a Microsoft Power BI<\/a> dashboard and download that data to Microsoft Excel<\/a>.<\/p>\n

Collecting data to solve problems<\/strong><\/p>\n

Because Maw was so close to the process, he saw the potential to solve problems through the trove of data the platform was collecting.<\/p>\n

\u201cLet\u2019s say that 80 percent of all requests get approved,\u201d Maw says. \u201cWe wanted to find out why that\u2019s the case. Tracking certain info allows us to update other aspects of our systems at the purchase order creation level, or even at the policy level.\u201d<\/p>\n

That data could then inform future optimizations.<\/p>\n

\u201cWe could make tweaks that would lower the number of requests made in the first place to 500 a month,\u201d Maw says. \u201cThen data from that version of the platform might give us what we need to get down to 300 or 200 a month. The idea is we just keep making the process better until the problem goes away.\u201d<\/p>\n

The impact of automation<\/strong><\/p>\n

The platform is already processing around 600 transactions per month globally, resulting in about 5,000 hours\u2019 worth of efficiency gains.<\/p>\n

Those gains come from a 30 percent reduction in the number of tickets and reduced errors stemming from automatic form validations. Automated routing also contributes by ensuring that only those who need to take action are looped in.<\/p>\n

Then there\u2019s the elimination of all those emails and the occasional Microsoft Teams call to nail down minor details.<\/p>\n

In addition, service-level agreements (SLAs) now take about three days to turn around, when previously they took twice that.<\/p>\n

\u201cWe saved days on SLAs,\u201d Busby says. \u201cLiterally.\u201d<\/p>\n

Surendranath echoed this statement.<\/p>\n

\u201cIt\u2019s improving productivity across the board,\u201d Surendranath says, \u201cincluding internal users, the Business Process Outsourcing team, and external suppliers.\u201d<\/p>\n

And because the entire process is now automated and centralized, nothing falls through the cracks, ensuring 100 percent compliance and leaving a comprehensive audit trail.<\/p>\n

In the near-term, the team plans to expand the scope of the platform to encompass a wider range of policies, and to modularize some of the components for reuse.<\/p>\n

But the real future of the platform lies in the actionable insights that are already being extracted from the data it collects.<\/p>\n

\u201cWe\u2019re close to 80 percent adherence to standard requirements already,\u201d Maw says, \u201cand of the remaining 20 percent, we know why 15 percent is non-standard, which is huge. That really narrows our focus on the remaining 5 percent.\u201d<\/p>\n

The team learned the ins and outs of Microsoft Power Automate by watching tutorial videos on the Microsoft Power Automate YouTube channel<\/a>. The DIY nature of the project was quite satisfying, Kelley says. \u201cThink about it. We built a full-featured enterprise app, complete with automation and reporting, without any heavy coding.\u201d<\/p>\n

It was also kind of fun, Busby explains.<\/p>\n

\u201cWe would brainstorm, and Korry and Binu would mock things up in a day or two, then come back and show us what the idea looked like,\u201d Busby says. \u201cWe saw our ideas come to life in a matter of days rather than weeks, and we could iterate from there. It was awesome.\u201d<\/p>\n

\"Related<\/p>\n