It’s no surprise Microsoft is at the forefront of intelligent forecasting. The technology leader leverages its own Azure Machine Learning service to improve forecast accuracy and drive business growth. The company also informs its financial plans with insights from external data.
So, how’s it going, and what can CFOs learn from Microsoft’s experience?
Victor Casalino, CFO for the Americas business at Microsoft, recently shared his thoughts in a webinar with Prevedere.
Here are some highlights of what Victor had to say.
Why insights from external data are valuable to CFOs
“The most strategic role that CFOs play is resource allocation. It requires a lot of courage and fortitude to move resources from one area of a company to the one that has the most potential. And to do that, you require fidelity in how you perceive the world. It requires a proper assessment of the market conditions, the economy, the needs of customers—so external data, as we call it.”
How external data informs business planning at Microsoft
“We use external data for a process that we call LRP, Long-Range Planning. Incorporating external data like that makes them part of the fabric of the forecasting process itself.”
Microsoft’s use of machine learning in forecasting
“Four years ago, we started to use this concept of machine learning. Machine learning for me is a huge econometric model. It harnesses the power of computing at scale. And it harnesses the power of being able to ingest all sorts of data. We use machine learning to inform our views of what the market is going to be one, two, three years down the road.”
The value of machine-learning forecasts at Microsoft
“We have found that having a machine-learning forecast solution as a guardrail to our business has improved our forecast accuracy and our ability to better guide resources quarter-in, quarter-out with analysts.”
Using machine-learning forecasts as a benchmark
“We use the outcomes of machine learning as a reference point. If you have a forecast from the field or a forecast from the engineering groups, we contrast them to what the models say. It’s not like you take the model blindly, because there might be elements that do require deeper inspection and discussion in the real world.”
The importance of an objective forecast in overcoming biases
“Invariably, sales leaders will use the forecast to manage the business for political means. People invariably derive value or personal prestige from the size of their budgets, from the size of the revenue that they manage. But you would have to be hard-pressed to disagree with a model that has proven to be consistently more accurate over time.”
The value of combining insight with conviction
“The ability to present data in a way that is compelling and easy to consume is extremely important. At the same time, what is required is the wherewithal and the courage to make the right calls. As a finance leader, you have to have the courage to see things for what they are, and to speak truth to power when the moment comes.”
How to get started
“If you’re changing significantly the way that you’re forecasting, you’ve got to start with a lot of focus so you start to show to the rest of the organization, the benefits and the power of that particular solution. We started with a consumer hardware business. Equally important is to make sure that you hand-pick the people who are going to lead this effort carefully. They should be curious and hungry for change.”
The importance of a CFO champion at Microsoft
“It does require the broad support of the higher levels of the organization. In our case, it was from Amy Hood (Microsoft EVP and CFO), where she basically said, ‛I’m not going to have a forecast discussion that does not include a solution that is a machine-learning forecast.”
Where Microsoft is today
“What we found over the quarters is that our models, our algorithms got better and better over time. We range from consumer hardware businesses, from Xbox to Surface devices, to Azure and Office 365, and we are using them across the board.”
Want to hear more from Victor?
You can listen to the entire webcast here.