Traditional forecasting vs. Connected Planning with machine learning

blog-MLvsPlanning-banners_1600x900

Today, there is a lot of enthusiasm in several tech segments about the tremendous advances in statistical forecasting, especially as it pertains to leveraging machine learning (ML). Generalized forecasting engines are making it easy for businesses to create automated forecasts to take advantage of their enriched data sets. This automation is helping take the guesswork and bias out of historically manual forecasting processes.

Yet, the question we are often asked is: “how do these forecasting platforms that leverage ML relate to Anaplan?

The short answer is that traditional forecasting is not business planning, nor is it necessarily better decision-making. A good forecast is important for informing decisions; however, it is usually just one component part among many that leads to sound business decisions for the future.

For example, let’s say you have a perfectly accurate forecast. Now what—what happens next? First, the forecast needs to be compared to what you originally planned or expected as an outcome. If there is a delta, you will need to decide a few things before acting:

  • Whether it is significant enough to react
  • If it is driven by an underlying change or just a temporary aberration
  • Whether it is because of an error in execution or in your plan’s assumptions
  • What are the fundamental changes that are not represented in the data that may undermine the ability to use historical data to predict the future?

To conduct sound business planning, you’ll need to put your plan in a broader context. That context will help you understand the drivers, dependencies, and sensitivities in the relationships between your forecast and the plans it informs. Taken together with your forecast, you can then consider how to react to the future.

Once you decide to act, it will likely trigger multiple related decisions where you will need to quickly model a revised business plan and build consensus on the actions and expected outcomes across various teams within your enterprise.

Let’s consider a quick hypothetical scenario. For example, what if we are in the business of running “the UBER of lemonade stands?” In such a scenario, we have pulled data from several sources into AWS Forecast—a strategic Anaplan partner platform that leverages ML, but does not require ML experience—to build an accurate prediction of pedestrian traffic and weather patterns. We have a fleet of independent drivers who will pull their hatchbacks up to target locations and sell the best tasting lemonade for the best possible price.

Then, our forecasting tool suddenly shows an unexpected spike in lemonade consumption in a specific city. We then bring that data into our city market model and see that it doesn’t look like the typical organic growth in a city. After some analysis of the data of where the lemonade was sold and when, we realize that we sold a lot of lemonade at a local NFL game (or hypothetically, a UEFA Champions League match in Europe). We further realize that the spike was unique because the Cleveland Browns (or AEK Athens Football Club) were in town playing at their home stadium, and since Browns (and AEK) fans have a predisposition to love the taste of bitterness, we have spotted a trend.

We now need to revise our production volume, supply chain and sales plans to incorporate the NFL or Champions League schedules, and the respective days and cities where these teams play on the road. We further have to build an alternative plan to acquire lemons, make lemonade, and distribute it to those markets in time for upcoming games.

Driver logistics need to ensure that we have more drivers on-call during home games to fulfill respective demands. Marketing also creates a campaign to put ads in Cleveland and Athens’ airports in the days leading up to road games to get their fans excited about more tasty lemonade.

All of those decisions are then modeled-out for our financial projections for future quarters, allowing us to raise expectations. Investors are happy, lemon growers are happy, and perhaps most miraculously, Browns and AEK fans are happy—all because we paired a better forecasting tool that leverages ML with the right planning and decision-making to properly respond to the forecast.

Forecasting is an important component to Connected Planning, and we are excited to continue working with leading innovation partners like AWS to bring the outputs of new forecasting and machine learning (ML) techniques into Anaplan’s platform. Indeed, the future is bright these days—even for Browns and AEK fans!

To learn more about more about Anaplan’s platform for Connected Planning, visit anaplan.com.

Related Posts

Blog Sign up

Share on facebook
Facebook
Share on google
Google+
Share on twitter
Twitter
Share on linkedin
LinkedIn
Share on pinterest
Pinterest