Microsoft Retains Its Finance Head Depend Flat With AI, Bots and Different Tech

A bunch of applied sciences, together with synthetic intelligence, bots, the cloud, information lakes and machine studying, are serving to Chief Monetary Officer

Amy Hood

hold a decent lid on finance head rely.

Cory Hrncirik,

who works on Ms. Hood’s workforce and leads Microsoft’s Fashionable Finance initiative, informed WSJ’s CFO Journal in regards to the new instruments, and why the group nonetheless makes use of Excel for some duties.

That is the primary a part of a sequence that focuses on how CFOs and different executives digitize their finance operations. Edited excerpts observe.

WSJ: When did Microsoft embark on its digitization journey?

Mr. Hrncirik: About seven or eight years [ago], we moved all of our information to the cloud. You need to take care of trying forward and making an attempt to grasp the way forward for your workforce. We name that technique and forecasting. We take into consideration the guide duties that we now have to do, and we take into consideration how we automate these. We centered lots about streamlining our information, creating one supply of fact.

WSJ: What’s the upside for finance staff?

Mr. Hrncirik: We wish to use expertise for areas the place [it] is suited to streamline and simplify the work that our individuals do. We wish them specializing in areas that, frankly, expertise nonetheless can’t assist us clear up very properly, like negotiating with enterprise companions or on the lookout for greenfield alternatives or managing advanced initiatives.

WSJ: How a lot is the corporate counting on machine studying?

Mr. Hrncirik: Our first foray into machine studying was within the forecasting area. Forecasting is one thing that each finance group does, no matter firm or group. For many, it takes a number of time. For many, it’s a number of heavy lifting in Excel, and it was for us as properly. Simply to place that in perspective, we usually would spend about three weeks each quarter constructing a forecast, and we might contain a thousand individuals in that course of, creating Excel spreadsheets in all of our subsidiaries and in all of our product groups. After which effervescent these forecasts up till they attain the CFO.

We launched machine studying again in 2015, and inside two quarters we realized that our algorithms weren’t solely performing in addition to the human-based course of, however we minimize our variance charge in half from about 3% to 1.5%. [Now], we will truly flip these fashions round in about half-hour.

WSJ: That is the quarterly forecast, appropriate? As an alternative of three weeks, that now takes about half-hour?

Mr. Hrncirik: That’s proper. We then push the insights out to our individuals round all of our subsidiaries. They nonetheless have an opportunity to have a look at them as a result of they convey distinctive information of native markets. They’ll typically say, “Oh, the all-up quantity seems to be good,” however we wish to alter among the seasonality or the break up between completely different merchandise or issues like that. Machine studying doesn’t all the time carry out rather well on the deep, granular degree.

WSJ: Are there different use instances?

Mr. Hrncirik: We’ve branched out and employed [it] in issues like compliance. We employed it in rushing up our inside audit course of. We make use of it in predicting recessions. We use it in our treasury group for analyzing paperwork from governments all over the world to grasp doable dangers. We use it even to determine which invoices may be automated and which want human intervention.

WSJ: What wanted to vary for that?

Mr. Hrncirik: Once I began my profession, I [had] to hook up with 50 completely different information units to drag data into Excel after which manually create insights from that information. We’ve moved all of these information sources, truly over 100 completely different [ones]. We’ve merged [them] in an information lake, and so that you merge all of that information collectively within the cloud. The second step is creating customary studies and analytical frameworks in order that we will discuss the identical enterprise the [same] manner all over the place all over the world.

WSJ: Are you utilizing bots?

Mr. Hrncirik: The usage of digital brokers was our foray into this world of synthetic intelligence. It’s pure language processing, both utilizing textual content or voice, the place this synthetic intelligence is just not solely understanding the phrases which might be spoken—in, by the way in which, over 60 languages—but in addition then inferring intent, and in addition streamlining a few of their dialog right into a thread. About 30% of 1,000,000 [internal] queries are dealt with completely by way of digital brokers now.

WSJ: The place are you deploying these bots?

Mr. Hrncirik: [For example] in our bill fee area. We course of 1000’s of invoices a month, and a number of these invoices are from the identical suppliers. What we present in doing that was that about 70% of all invoices may very well be automated. We educated a machine-learning algorithm to truly discover the 70% and simply pay them.

The algorithm additionally says, “Hey, by the way in which, we’ve detected an irregularity or an anomaly on this geography, or on this particular [stock keeping unit] or product space.”

WSJ: How correct is the expertise?

Mr. Hrncirik: Our error charge has gone from about 2% to lower than 1%. It’s so correct since you don’t have people manually coming into information into the system or manually doing among the calculations and different issues.

WSJ: Will these instruments let you carry down your finance head rely?

Mr. Hrncirik: For those who truly have a look at most finance groups, the pinnacle rely grows in lockstep with enterprise progress, and that was the case for Microsoft. All through our historical past, from the ’70s, ’80s and ’90s, as we added extra income, we added extra individuals.

The downturn of 2008 and 2009 was a catalyst for us. There was a call made right here at Microsoft to maintain our [finance] head rely flat. We’ve now achieved that over the previous decade. Over the identical timeframe, our income has [nearly] tripled. Clearly, our market cap now could be over $2 trillion, and the enterprise is rather more advanced, and but we now have [roughly] the identical variety of individuals [in finance].

WSJ: Is the finance group nonetheless utilizing Excel?

Mr. Hrncirik: We love Excel, and we use it typically. Excel has a spot and all the time could have a spot.

Write to Nina Trentmann at Nina.Trentmann@wsj.com